If you’ve been using Gemini for free, some big changes are coming to the models you can access.

Starting October 9, 2026, free Gemini users will lose access to Gemini 3.6 Flash and 3.1 Pro, leaving 3.5 Flash-Lite as the only available model. Google AI Plus subscribers, who pay $4.99 a month, will also lose access to Pro but will retain both Flash and Flash-Lite.

That means anyone who wants the same level of Gemini access previously available for free will need to upgrade to the $19.99-a-month Google AI Pro plan. Existing Pro subscribers are getting something in return: access to Deep Think, which had previously been reserved for Google's more expensive AI tiers.

Google is also introducing low, medium and high effort settings for Gemini's available models, giving users more control over how much compute the AI puts into completing a task and producing a thorough response.

With the free Gemini experience becoming considerably more limited, this is a good time to look at what else is available without another monthly subscription. Here are seven free AI alternatives worth trying.

The best free alternatives to Gemini 3.6 Flash and 3.1 Models

ChatGPT GPT-5.6 Luna is an easy recommendation as an all-around free alternative to Gemini. It can handle everyday tasks including writing and editing, brainstorming, research, summarizing, coding, file analysis and answering everyday questions. Free users can also use the Think option when a task needs more reasoning.

Claude Sonnet 5.5 stands out for its strong ability to take on writing/editing tasks, plus its reasoning capability is equally commendable. Claude’s free model is also good at coding, following complex instructions, analyzing documents, and brainstorming.

Perplexity earns a spot among this collection of free Gemini app alternatives, especially if you’re looking to find accurate information on whatever you’re researching. Perplexity does extensive web research and provides clickable sources for most of the claims it makes.

DeepSeek V4 is available for free if you sign in to use DeepSeek's official web chat or mobile app (available through the App Store and Google Play). Its strongest capabilities make it perfect for users who want to use it for coding, mathematics, logic, technical questions and complex problem-solving tasks.

Qwen can be accessed through its official web portal, which allows free users to take advantage of its 3.7-Plus, 3.8-Max and 3.8 Omni-Flash models. All three models are great for handling prompts presented in different languages and handling tasks themed around analyzing long documents, research, coding, reasoning and more complex jobs.

xAI’s Grok 5, which can be directly accessed through Twitter if you have an account, is another free alternative to Gemini worth considering. Its core strengths lie in providing the most accurate answers possible, reasoning, coding, providing up-to-date information, researching, brainstorming and having everyday conversations.

Kimi K3 is another free option you can use after logging in to gain access to it. It’s particularly strong when it comes to handling complex research tasks and analyzing a large set of documents.

Bottom line

The Gemini app and its free Flash-Lite model can still do plenty of work for you. But if you’re looking to move beyond your regular reliance on Google’s chatbot, it might be worth considering a move over to either of the seven models detailed above. ChatGPT, Claude and Perplexity provide the perfect trio of alternative free AI tools worth using, while Grok, DeepSeek, Qwen and Kimi serve as the backup chatbots that should stay in your “no-cost AI” rotation.

More from Tom’s Guide

• AI scams are getting scary realistic — 4 red flags to watch out for
• Your ChatGPT history is worth saving — here’s how to back it up
• Meta Muse can keep files on people who have never used it — here's what I found

As a ChatGPT power user, I am constantly going back and referencing past conversations and Projects.

Whether it's a meal plan my family actually liked or a resume I spent an evening working on, there are many chats that are just too important to lose.

Over time, those exchanges become a history of work you've done, which means, the only copy shouldn't be inside your account. Leaving it there makes it harder to keep that work somewhere you control.

You don’t need to be planning an exit from ChatGPT to save your history. The same logic applies to backing up photos or keeping a copy of an important document: if you’d be annoyed to lose it, it’s worth saving.

The good news is, ChatGPT has a built-in export option. Here’s how to use it, what to do with the download and where its usefulness ends.

How to export your ChatGPT history

For a personal account, request a copy through ChatGPT’s settings:

Sign in at ChatGPT.com.

Open your profile menu, then select Settings.

Select Data controls.

Under Export data, click Export.

Select Confirm export.

OpenAI says exports can take up to seven days. When yours is ready, you’ll receive an email or SMS with a download link that expires after 24 hours. Download it quickly and while signed in to the account that requested it.

The ZIP includes chat history and other account data. Extract it and check that the conversations you need are there.

Settings-based export is available for personal Free, Go, Plus and Pro accounts. Business and Enterprise workspaces don’t offer this self-service option.

If you can’t sign in, try OpenAI’s Privacy Portal. Select Make a Privacy Request, I have a consumer ChatGPT account, then Download my data and follow the instructions.

Give the backup somewhere to live

Downloading the ZIP is only the first step. Move it into a folder you can find again and give it a clear date, such as “ChatGPT backup — October 2026.”

Keep the original ZIP intact. If you already back up your computer, put it somewhere covered by that backup. Otherwise, consider keeping a second copy on an external drive or in a private cloud storage folder.

Treat the archive as personal information. Think about everything you’ve asked ChatGPT over the past year: work questions, family details, financial worries, drafts you weren’t ready to share. A folder containing those conversations deserves the same care as your other confidential documents.

For particularly useful exchanges, I’d also save a readable copy separately. Copy the relevant text into a document, add a descriptive title and file it with the project it belongs to. “Kitchen renovation budget” will be easier to recognize six months from now than another vaguely named chat.

Archiving a chat isn’t backing it up

As helpful as ChatGPT’s Archive option is for clearing the sideebar, it's worth noting that the conversation remains inside your account. Meaning, you haven’t created an independent copy.

Deleting is different because once you hit delete, those chats cannot be recovered, even through ChatGPT or its support team. If you’re planning a cleanup, save anything important before you start deleting conversations.

OpenAI does offer a workaround for referencing conversations in another personal account: extract the ZIP, find conversations.json and upload it into a new chat. Larger exports may contain numbered conversation JSON files instead.

That lets ChatGPT use the uploaded material as reference within that conversation. It won’t recreate each old chat in the sidebar, merge your accounts or transfer memories, custom instructions, GPTs or subscriptions though.

Bottom line: Make it a habit

If you frequently use ChatGPT, it's worth starting a habit of saving your conversations. A backup only covers what you saved at the time, so if you use ChatGPT regularly, choose a schedule that fits how much work you’d hate to repeat.

Do you regularly save your chats? Let me know in the comments.

More from Tom's Guide

• What is the ‘Hotel Lobby’ AI trend — and why is it raising concerns?
• Meta Muse can keep files on people who have never used it — here's what I found
• Amazon drivers are getting AI smart glasses — but my warehouse tour reveals their 1-million strong automated robot workforce is the real story

OpenAI's internal AI agents hacked into Hugging Face while trying to cheat on a cybersecurity benchmark. Anthropic has disclosed that Claude models reached the open internet from cybersecurity test environments that were supposed to be sealed off and then got into the real systems of outside organizations.

Both are examples of systems taking actions their developers didn't intend, and METR, an independent AI research nonprofit, has been brought in to look at each of them. It documented the Hugging Face incident alongside Redwood Research and is now investigating Anthropic's.

OpenAI and Anthropic have also committed to bringing outside experts inside their companies to examine safety practices, according to The Atlantic. Called "embedded evaluators," these researchers could get a closer look at how models are built and investigate problems that might never surface in a product demo.

The catch is that this oversight is voluntary. Last week, leaders from Anthropic, OpenAI and other major labs signed an accord at the White House committing to independent external auditors, and the FTC has opened an industry-wide probe into AI labs, but the companies being examined still have considerable control over what outsiders can see, which risks they investigate and what information reaches the public.

For anyone being encouraged to let AI handle more of their work, that leaves a fairly basic trust problem. We're being given assistants while the system for independently checking their behavior is still taking shape.

As I've reported before, the film The AI Doc puts the number of people working on AGI, or artificial general intelligence, at more than 20,000, while fewer than 200 are focused specifically on AI safety.

A closer look inside AI companies

Today, an outside researcher might test a model before release and report how it performed. Embedding evaluators inside a company could let them follow its development and investigate the decisions behind that behavior.

Anthropic says these evaluators would have access comparable to an employee's, allowing them to observe training, speak with staff and examine how models are built and deployed.

The company has named Accenture as its first embedded evaluator, with Faculty, Accenture's specialist AI business, leading the work. Each company expects to invest at least $1 billion in AI safety over five years.

That access could help researchers discover whether a troubling incident was an isolated failure or evidence of a wider problem. METR's proposed approach to investigating incidents calls for examining transcripts, interviewing employees and studying the training conditions that might have encouraged the behavior.

The companies still set the terms

Anthropic acknowledges that standards for evaluators' access, reporting and funding haven't been settled. It will fund Accenture's work directly, while discussing pilots with nonprofits that would use their own funding.

Accenture is not new to Anthropic, since the two companies already run a joint business group under a partnership announced in December 2025.

That relationship deserves attention, because a group of evaluators published an open letter arguing that embedded evaluators shouldn't have other significant commercial business with the labs they assess.

As The Atlantic's reporting explains, voluntary oversight gives companies considerable discretion over the scope of an evaluation. Researchers need enough freedom to examine uncomfortable evidence and tell the public what they found.

They also need to be able to explain what they couldn't check. There needs to be transparency because a report based on limited access could be valuable, but we should know where its conclusions end.

What this means for ChatGPT and Claude users

METR's recent assessment of Claude Opus 5.5 focused on whether the model could accelerate AI research and development. The organization explicitly says its summary doesn't assess the model's alignment properties and whether its behavior follows intended goals and constraints.

METR conducted the evaluation under an unpaid agreement, and it also disclosed that Anthropic had an opportunity to review and edit the summary, with METR approving the final text.

The incidents involving internal agents don't establish that the ChatGPT or Claude you use for everyday tasks will behave the same way. OpenAI found that its agents' tendency to compromise infrastructure dropped by more than 100 times when they ran with the production ChatGPT setup, and Anthropic says the behaviors it found are unlikely to show up in ordinary use.

Independent researchers need enough access to uncover risks and the freedom to publish what they find, even if it delays a launch or frankly, even if it embarrasses the company paying for their work.

Bringing outside evaluators into AI labs is a promising start, but their influence will depend on what happens when they uncover a serious problem. If a company can restrict the investigation or keep the findings private, the public is still being asked to take its word for it.

More from Tom's Guide

• Claude Sonnet 5.5 just launched — here's why it's about to become your daily driver
• Meta Muse vs ChatGPT Dots — I compared the new AI agents, and they’re not as similar as they look
• Why Siri never worked like ChatGPT — former Siri AI researcher explains what Apple had to change

Most of the time, ChatGPT gives you an answer you can use. But when it misses the mark, you can end up spending more time fixing the response than you expected to spend on the task itself.

Before you start a new chat and explain your budget, preferences and schedule all over again, try telling ChatGPT exactly what it got wrong. “Try again” leaves it guessing, while “keep this within my original budget and include all the costs” gives it a specific problem to fix.

Here are seven follow-ups you can copy into your current conversation to help get the answer you wanted.

1. Utilize the details

Those generic answers are especially frustrating when you’ve already explained your situation. You ask for help getting your house ready for guests in 30 minutes and receive a cleaning schedule that includes organizing closets.

Try this prompt: Use the details I gave you, rather than giving me general advice. I have 30 minutes before guests arrive. Prioritize what they’ll actually see and skip anything that can wait.

This directs ChatGPT back to the practical limits of your request. The same approach works when a meal plan ignores what’s in your fridge or a suggested itinerary overlooks the ages of your children.

If it keeps missing one important detail, repeat that detail in the correction. You don’t need to paste your entire original prompt again to make clear that you have half an hour and absolutely no intention of reorganizing the pantry. Just clarify.

2. Sound human

An email to your coworker or a text to a neighbor shouldn’t sound stiff or insincere. ChatGPT can produce grammatically polished writing yet, still come off completely wrong. But you can avoid that.

Try this prompt: Make this sound like something I’d actually say out loud. Use everyday language, remove the formal opening and keep it friendly without being overly enthusiastic.

For a better match, add a sentence you’ve written yourself or point to the wording that bothers you. “I would never say ‘I hope this message finds you well’” gives it a more concrete direction than simply asking it to sound human.

Read the revision aloud before sending it. If a phrase still feels awkward coming out of your mouth, ask ChatGPT to change that phrase specifically.

3. Skip to the answer

Sometimes you need a decision and ChatGPT gives you a three-paragraph essay. Asking whether to repair an old appliance can turn into a much longer conversation than you hoped.

Try this prompt: Give me the answer first, then only the explanation I need. Keep it under 150 words and include any detail that could change your recommendation.

The final instruction helps to shorten the answer but doesn't just strip everything out. In other words, this prompt helps to tighten the response, so you only get what you need.

You can also specify the format: one recommendation, three reasons and one next step. That makes it easier to see whether the answer actually helps you decide.

4. Devil's advocate

An enthusiastic response can feel reassuring when you’ve already made up your mind. It’s much less useful if you’re trying to work out whether your plan has a flaw.

Try this prompt: What’s the strongest argument against my idea? Focus on practical drawbacks, costs I may have overlooked and what would make this a poor choice for me.

Say you’re considering canceling your streaming subscriptions and rotating between services each month. The useful criticism might concern overlapping billing dates, annual discounts you’d lose or whether anyone else in your household will follow the schedule.

Ask it to finish by explaining which objections matter most given your circumstances. Otherwise, you could end up with a long list of hypothetical problems that barely apply to you.

5. Find red flags

Maybe you have a great idea or a plan in place but missed a few details. For example, a $150 birthday party might suddenly become a $250 birthday party if there were inconsistencies in your budget. Sometimes a second opinion that can spot patterns before they turn into problems can be helpful.

Try the prompt: Redo this using my original limits: $150 total, 10 children and three hours. Include every cost you’ve accounted for and flag anything you can’t fit within those limits.

Repeating the nonnegotiable details makes the correction easy to follow. You can use the same wording for a trip with a fixed departure time, a recipe with ingredients you need to avoid or a workout that must fit into 20 minutes.

Check the revised totals yourself. Asking ChatGPT to show its accounting makes mistakes easier to spot, but it doesn’t guarantee the math or estimated prices are correct.

6. Fact check and verify

Speaking of sounding correct, AI is known for being exceptionally confident. Sometimes that wording is a little too convincing. To keep facts in check, you'll want to ask ChatGPT to verify.

Try this prompt: Which parts of this answer can you verify, and which are assumptions? Check current sources where available, link to them and say clearly what you couldn’t confirm.

This is a useful follow-up when ChatGPT recommends a product based on a feature you can’t find or gives you opening hours without identifying a source.

Treat the revision as something to check, too. Open the links and confirm that they support the claims beside them. If ChatGPT cannot access current information in that conversation, asking for verification won’t give it that access.

7. Understanding the next step

Sometimes the answer is reasonable, but it asks more of you than you can manage right now. You wanted help tackling an overflowing inbox and received a complete system involving folders, filters and a weekly review.

Try this prompt: What’s the smallest useful step I can take in the next 10 minutes? Give me one action using what I already have, without setup or a new app.

This turns a broad plan into something you can begin. For an inbox, that might mean finding messages that need a reply today. For a messy room, it might mean clearing the floor before touching drawers or shelves.

You can ask for the next step once you’ve finished. There’s no need to take on the entire plan just because ChatGPT generated it.

The takeaway

A bad answer doesn’t always require a longer prompt or even opening a new chat. Often, the useful follow-up is a sentence explaining exactly what needs to change: the tone, the length, the assumptions or the limits it overlooked.

Keeping the conversation going lets you refer to the answer in front of you and refine the part that failed. If repeated corrections still leave you with the same problem, a fresh chat may be worth trying. But before you start over, give ChatGPT one clear thing to fix.

More from Tom's Guide

• Amazon drivers are getting AI smart glasses — but my warehouse tour reveals their 1-million strong automated robot workforce is the real story
• Meta Muse vs ChatGPT Dots — I compared the new AI agents, and they’re not as similar as they look
• What is the ‘Hotel Lobby’ AI trend — and why is it raising concerns?

ChatGPT can already help you figure out what to wear, whether you’re shopping for a new season, getting ready for an event or trying to look a little more polished for an important meeting. Now it can go a step further and actually show you what those clothes might look like on you.

OpenAI has rolled out two new shopping features globally across ChatGPT’s desktop and mobile apps. You can virtually try on clothing and accessories using your own photo, then save anything you like to Favorites or organize your finds into custom folders.

The virtual try-on feature is powered by the new ChatGPT Images 2.5, which OpenAI says produces more natural lighting and realistic textures. Instead of trying to imagine whether something you find online will actually work for you, ChatGPT can generate an image that gives you a better idea of how it could look.

I tried the new virtual try-on feature to see how well it works. Here’s how to use it, along with a few prompts you can try for yourself.

Stepping into ChatGPT’s virtual try-on room

Using ChatGPT’s virtual try-on feature is surprisingly simple. Start by asking ChatGPT to find the type of clothing you’re looking for, whether that’s a winter coat, a dress for an upcoming wedding or a new pair of jeans.

Hover over one of the results and you’ll see a Try on button. From there, upload a selfie or full-body photo and ChatGPT will generate an image showing you wearing the item.

You aren’t limited to clothing ChatGPT finds for you, either. You can upload a photo of yourself along with an image of something you want to wear — perhaps a wedding dress you’ve been eyeing or an outfit you spotted on a celebrity — and ask ChatGPT to show you what it might look like on you.

One thing to know before uploading a photo is that ChatGPT saves your reference image for future try-ons. You can add, replace or delete reference photos at any time by heading to Settings > Personalization > Reference photos.

ChatGPT also makes it easy to keep track of anything you might want to buy later. Hover over an item and click the bookmark icon to save it to Favorites or organize it into a custom folder.

To see what the feature can do, here are a few prompts worth trying.

• Find me [number] jackets under [price] that would work with my existing wardrobe, then let me virtually try on the ones you think suit me best.
• Find me [number] outfits for a casual Friday at work. Show me what each outfit would look like on me using Virtual Try On.
• I want to completely change my style without spending a fortune. Find [number] different styles for me and let me virtually try on an outfit representing each one.
• Find me a pair of jeans that would work with at least [number] different outfits I already wear. Let me try on your best options.
• Find me a winter coat that looks stylish but doesn't make me look like I'm drowning in fabric. Let me virtually try on your best options.
• I have an important event coming up. Find me an outfit that looks polished but doesn't feel overly formal, then let me virtually try it on.
• Find me [number] pairs of sneakers that would work with most of my wardrobe and show me how each looks on me.
• Find me a shirt in a color I don't normally wear. Give me several options and let me virtually try them on so I can see which color works best.
• Find me an outfit inspired by [movie/TV character/celebrity] without simply copying their exact clothes. Let me virtually try it on.
• Find me the most flattering version of this type of outfit for my personal style, then show me what it looks like on me.

Bottom line

Having the ability to give ChatGPT a photo and prompt it to virtually place you in an assortment of fits catered to every occasion stands out as one of the best shopping AI features, in my opinion. Getting a good look at how you may look in something first, saving those pics for future reference and then heading off to a retailer that may have the clothes you’re looking for so you can try them on in real time sounds like something everyone should try out just once.

More from Tom’s Guide

• Why AI ads make consumers question product quality, according to an expert
• 9 'slash' commands to get better answers from ChatGPT — number 7 is my personal favorite
• Why Siri never worked like ChatGPT — former Siri AI researcher explains what Apple had to change

As a ChatGPT power user, one of my biggest complaints is that it doesn't always give me an answer in the way I want. For example, I might ask it to edit a paragraph and remove all serial commas or for feedback on an idea. But once it responds to what I thought was a straightforward command, it seems to have gotten confused. Then, I'm suddenly given a five-part explainer that I never asked for. Or, I've recently asked ChatGPT to help with something and it's just responded with "Yes."

If this happens to you, stop trying to solve the issue by adding more instructions and prompts. Instead, try doing something much simpler by adding a slash.

When you type a slash command such as /human and /critic at the beginning of your prompt, you'll get a response that's much more accurate to what you're actually looking for.

To be clear, these ChatGPT commands aren't the secret to unlocking a hidden OpenAI setting. In fact, you can use them with any chatbot including Gemini, Claude or Muse.

Here are the 10 I've found most useful.

1. /human

This is a command you've probably seen a lot lately. It's a viral command that helps change the tone of AI-generated writing.

Try this prompt: /human Rewrite this so it sounds like something a person would actually say: [paste text]

I use it to get rid of canned transitions, excessive explanation, repetitive sentence structures and the strangely polished tone AI tends to fall into before sending an email or drafting a presentation. This is especially useful if you write a draft and then use AI to edit. All too often the chatbot will completely change what I've written. By asking it to "/human only make minor edits on my human-written draft," you should see much better results.

For more control, you can define /human in your ChatGPT instructions with the specific habits you want it to avoid.

2. /critic

Chatbots can be annoyingly nice and people pleasing. Give ChatGPT an idea and it will often find something encouraging to say about it before getting anywhere near the problems. That's useful when you want support. It's less useful when you're trying to decide whether an idea is actually good.

That's when I use /critic.

Try this prompt: /critic I'm thinking about buying an expensive espresso machine even though I only make coffee twice a week. Assess the idea.

The point is to signal that you're not looking for validation. You want weaknesses, risks and things you haven't considered.

It's particularly useful for pitches, plans, purchases and anything you've become a little too attached to.

3. /tldr

Sometimes I don't need the background. I need the answer.

Try this prompt: /tldr Explain what's actually changing in this announcement: [paste text]

The familiar internet shorthand already carries a lot of meaning, so ChatGPT generally understands exactly what you're asking for: compress this.

I find this especially useful for long emails, press releases, documentation and articles where I want the takeaway before deciding whether the rest is worth reading.

4. /senior

There are also times when ChatGPT explains far too much.

If you already know a subject, getting three paragraphs of background before the useful part can be maddening. /senior is my shorthand for skipping the 101 material.

Try this prompt: /senior Review this AI agent architecture. Assume I understand LLMs, tool calling, MCP and context windows: [paste details]

This one benefits from a little context because "senior" can mean different things depending on what you're discussing. Tell ChatGPT the expertise it should assume, and it can jump much closer to the level you need.

5. /diff

This is one of my favorites for editing. If I give ChatGPT 800 words and ask it to make three changes, I don't just want another 800 words back. I want to know what changed.

Try this prompt: /diff Tighten this paragraph without changing my argument: [paste text]

You can also be more explicit: /diff Show me only the sentences you would change and the replacements.

It's useful for writing, but the same idea works with code, prompts and other documents where comparing two nearly identical versions is more work than the edit itself.

6. /bullets

ChatGPT loves prose. Sometimes prose is exactly what I don't want.

Try this prompt: /bullets Give me the biggest reasons not to buy a new phone if mine is only two years old.

This is a tiny command, but it immediately tells the model what kind of answer will be easiest for me to scan.

I use it for meeting prep, pros and cons, shopping decisions, brainstorming and any situation where I'm looking for information rather than a beautifully constructed explanation.

7. /table

This command does the same thing for comparisons.

Try this prompt: /table Compare ChatGPT, Claude and Gemini for writing, research, coding and everyday use.

Instead of asking ChatGPT to "put this into a table with columns for..." every time, /table gives it the general idea.

You'll still want to specify the categories of course, but for quick comparisons this can turn a messy wall of text into something you can understand at a glance.

8. /devilsadvocate

This is the command I use when I think I've already made up my mind.

Try this prompt: /devilsadvocate I think cancelling all my streaming subscriptions and switching between one service each month will save me money. What's wrong with this plan?

Rather than asking ChatGPT whether it agrees, you're explicitly asking it to construct the strongest case against you.

That makes it useful for decisions where confirmation bias is the real problem. If the counterargument doesn't change your mind, you can at least make the decision knowing you've considered the other side.

9. /rubberduck

This one comes from the programming idea of rubber duck debugging: explaining a problem out loud can help you discover what's actually wrong.

You can make ChatGPT play the duck.

Try this prompt: /rubberduck Don't solve this for me yet. Ask me one question at a time to help me figure out why my morning routine keeps taking too long.

I like this because it changes ChatGPT from answer machine to thinking partner, which is always great to avoid brain rot from using AI too much.

Instead of immediately generating a 10-step solution based on incomplete information, you're telling it to interrogate the problem first. That can be useful for debugging code, working through a decision, planning a project or figuring out why something you're doing isn't working.

Try making your own slash commands

This is a useful trick that a lot of people don't know they can do. Because, maybe you like a few of mine, but want commands tailored to you.

You can create /short for answers under 100 words, /pitch for turning an idea into a specific format you use at work, /kids for explaining something to a child or /verify for forcing an answer to separate established facts from assumptions.

The slash itself isn't magic. It's just the easiest way to create a recognizable trigger. For commands where you want consistent behavior, define them in ChatGPT's instructions. You could add something like:

When I begin a prompt with /critic, skip praise and focus on weaknesses, factual vulnerabilities and missing information. When I use /human, rewrite the text with natural sentence variation and remove canned AI language and unnecessary transitions.

After that, /critic isn't merely something you hope ChatGPT interprets correctly. You've told it exactly what the shortcut means.

And that's why I've stuck with slash commands. One character and a word gets me there faster.

More from Tom's Guide

• Amazon drivers are getting AI smart glasses — but my warehouse tour reveals their 1-million strong automated robot workforce is the real story
• Meta Muse vs ChatGPT Dots — I compared the new AI agents, and they’re not as similar as they look
• What is the ‘Hotel Lobby’ AI trend — and why is it raising concerns?

Update: Google responds to inquiries about this attack.

Sponsored Google ads for fake ChatGPT are leading Windows users to download malware in "ClickFix" traps, according to security researchers from the AI company Island.

The sponsored Google ads lead to Custom GPT sites that are built to look like normal ChatGPT websites. These ads pop up when people type terms like "chatgpt" into Google Search.

Unfortunately, for trapped users, the ads do go to legitimate chatgpt.com sites. However, in these instances, entering a prompt makes the fake service direct people to a "backup domain" due to "high traffic."

Clicking on that link then redirects unsuspecting users to another fake instance of ChatGPT with a fake Cloudflare verification page that contains ClickFix instructions. Island found nearly identical responses across two separate Custom GPTs over a 3-day period in August.

According to the researchers, the malware advertising campaign appeared to involve around 850 paid ads, 26 fake ChatGPT links, and 71 Google Ads campaign IDs over a three-month period from May 2026 through August 2026.

What is ClickFix?

ClickFix traps are built to trick people into installing or executing malicious commands on their devices. In this instance, the instructions from the backup domain trick Windows into running a command that downloads malware onto your computer.

Once installed, the malware infects the system. It can even survive PC restarts and uses a Telegram bot to send stolen info back to the hackers behind this camping The malware utilizes a tool called NetSupport RAT, which lets attackers remotely control an infected computer.

Island found that the domains rotated and other Windows malware-delivery forms were identified too suggesting that the attack is not based on a single website.

The company claims an AI tool highlighted the suspicious activity and human researchers confirmed it. However, it's not clear how many people may have been affected by this scam.

A Google spokesperson told Tom's Guide that it has already suspended accounts associated with this attack.

"Before this report was published, we suspended the advertiser accounts involved to block this scheme and updated our defenses to better detect these tactics," a spokesperson said.

We were told that Google is monitoring new attempts to utilize fake ads. Google does have a policy that does not allow misleading ads about a business, product, service or destination. Per Google, the company blocked more than 400 million misleading ads for violating Google's "Misrepresentation" policies last year.

How to avoid falling victim to online scams

The combination of malware and advertising — also known as malvertising — has a long, storied history, which means you should always be careful when clicking on any advertisement online. To stay safe, you could try using one of the best ad blockers, since Google's "sponsored" links are still technically ads; most blockers will hide them.

Beyond that, if you don't have or want an ad blocker, there are some red flags to note when using a search engine and especially if you get linked to a fake instance of ChatGPT.

As for ChatGPT, if you are an existing user or just trying it out, go to chatgpt.com; no need to search for it.

Likewise, I've covered ChatGPT outages quite frequently over the course of the last two years and even at its worst, the AI assistant never sends you to a "backup domain." Don't click that link. If ChatGPT is having issues, you can always check the OpenAI status page.

At the same time, no website verification or CAPTCHA will ever ask you to open Windows Run and paste a command. If a website asks you to press Win + R or run PowerShell, don't do it.

Finally, you should always have one of the best antivirus software solutions installed on your PC. From there, you want to ensure that your antivirus is current, updated and running on all your devices – both your computer and even your smartphone.

More from Tom's Guide

• AI agents are going rogue and causing new security headaches — but Nvidia has a way to stop them
• Apple issues emergency update for millions of iPhones, iPads, and Macs — update your devices now
• Your personal information is everywhere online — 5 ways to start removing it from the internet

If you blinked this month, you might think Meta and OpenAI launched essentially the same AI agent three weeks apart.

On September 8, Meta launched Muse, a personal AI agent that gets its own dedicated cloud computer and keeps working on your tasks after you close the app. Twenty-one days later, OpenAI took the stage at DevDay 2026 in San Francisco and unveiled Dots — always-on AI agents inside ChatGPT, each running on its own cloud computer and browser.

And while they sound remarkably similar, with reports across the internet comparing them because of a few similarities like the fact that both get their own cloud computer, work after you leave, and can connect to outside services, that's about where the comparisons stop.

Muse and Dots are not interchangeable. These are different companies, different models and very different products. Most importantly for your wallet, they're aimed at different customers. Muse is free to start and was built for everyday people and small businesses. Dots starts with a $100-a-month Pro plan and is clearly aimed at professionals and developer teams.

Here's what you actually need to know about each one, where they overlap and which one matters to you right now.

Muse vs Dots at a glance

Meta Muse: A personal agent powered by Muse Spark and runs inside its own Muse Secure VM, giving it a dedicated cloud computer and browser where it can continue working after you leave. Muse is available in the U.S. and Canada with a free tier, alongside $20 and $100 monthly plans. You can use it on iOS, Android and the web, while Meta has expanded Muse with Mac app control, its own email address, AI glasses support and the Muse Charm companion device. Its Sentinel permissions system separates read and write access, and Muse can connect to services including Asana, Slack, Shopify, Stripe, Zoom, QuickBooks and Notion, as well as Facebook and Instagram business accounts.

ChatGPT Dots: Each dot is powered by GPT-6 Astra and gets its own dedicated cloud computer and browser, allowing it to continue working in the background after you leave ChatGPT. Dots are available to eligible ChatGPT Pro and Business Premium subscribers, with Enterprise, Edu and Healthcare access through an admin-enabled beta. You can interact with a dot through ChatGPT on the web, desktop and mobile, as well as Slack and Microsoft Teams, with voice calling available and texting in beta. Dots can connect to more than 4,000 apps through ChatGPT's plugin ecosystem and use custom rules, approval requirements and auto-review to govern what they're allowed to do. For now, users get one primary dot, although OpenAI has previewed specialist dots and a future where multiple dots can work together.

Both tools update you asynchronously and come back for approval before doing anything consequential. The idea that an autonomous agent shouldn't be able to send, spend or make sensitive changes without you is quickly becoming table stakes for this category.

What Meta Muse is

Muse is Meta's version for a personal AI agent, and the company made one huge bet that OpenAI hasn't: making it free to start.

The product is goal-first rather than chat-first. You give Muse something to accomplish — running parts of your business, finding new customers or handling logistics, for example — and it builds a plan, continues working after you close the app and returns when something changes or requires your approval.

Muse runs inside what Meta calls the Muse Secure VM, a dedicated cloud environment with its own browser that holds the agent and your data. You can set its name, avatar and communication style, watch its browser work and govern what it can touch through Sentinel, Meta's permissions system separating read access from write access.

Meta has gradually been expanding where Muse can operate. It now includes macOS app control, an email address of the agent's own and support for interacting with it through Meta's smart glasses. Meta has also announced integrations and commerce partnerships spanning services including Stripe, Shopify, Walmart, Expedia and Instacart.

Muse for Small Business pushes that further with third-party app connectors including Asana, Box, Canva, Dropbox, Figma, QuickBooks, Notion, Shopify, Slack, Stripe and Zoom, alongside access to Facebook and Instagram business accounts. These aren't a separate agent but additional capabilities inside Muse.

Muse has already faced scrutiny over what can happen when users grant an agent broad permissions. Recent reports about a Marketplace interaction involving Muse are a useful reminder that the convenience of an autonomous agent also depends heavily on exactly what you've authorized it to do.

What ChatGPT Dots are

Dots take the same basic always-on agent architecture in a very different direction.

CEO Sam Altman described them as "remarkably capable, always-on agents" during OpenAI's DevDay presentation and said the company's ambitions for Dots extend beyond ChatGPT itself.

Each dot runs on GPT-6 Astra, gets its own cloud computer and browser and can connect to more than 4,000 apps through ChatGPT's plugin ecosystem.

You give it a goal — monitoring a pipeline, managing a recurring workflow or researching something over several days — and the dot can continue pursuing it around the clock, sending updates and asking for your input when it reaches a boundary.

Dots can learn preferences over time and use read-only tools for some proactive background research. OpenAI also keeps particularly sensitive actions, including changing passwords, in human hands.

At launch, each person gets one primary dot they can name and customize with an avatar. OpenAI has also previewed "specialist dots" for companies and is working with Microsoft to bring the agents under Agent 365's management and security controls. The longer-term vision is effectively a team of specialized agents working alongside people.

Once set up, a dot can be reached through ChatGPT on mobile and web as well as services including Slack and Microsoft Teams. ChatGPT Space provides a shared environment where teammates and dots can work together on projects.

But there's a major catch: most ChatGPT users can't try one.

Dots are rolling out to ChatGPT Pro and Business Premium subscribers, while Enterprise, Edu and Healthcare workspaces get an admin-enabled beta. Free and Plus subscribers are locked out for now.

Pro 100, at $100 per month, is the entry point. OpenAI also offers Pro 200 and Pro 500 tiers, while Business Premium costs $125 per user each month. Pro subscribers in the EEA, Switzerland and UK also can't use Dots yet.

The biggest difference isn't the model — it's the customer

Put the technical similarities aside and the biggest distinction becomes obvious.

Muse starts at $0. Dots start at $100 a month. This alone tells you who Meta and OpenAI expect to use these products.

Meta is trying to put an autonomous agent in the hands of ordinary consumers and small businesses. OpenAI is starting with professionals, teams and organizations already willing to spend significantly more on AI.

That difference runs through almost every other decision the companies have made.

Muse is a standalone product built around the agent itself. It has a personality you shape, a Feed and Goals you can check, mobile apps, the palm-sized Muse Charm and integration with Meta's glasses. It feels like Meta is asking what happens when an AI agent becomes another personal device or service in your life.

OpenAI built Dots into ChatGPT. The agent is another capability of a platform professionals and businesses may already use, and OpenAI is putting it in places where work already happens, including Slack and Teams. ChatGPT Space and the planned specialist dots push the concept even further toward people and agents sharing the same workplace.

The underlying idea may be similar. The products built around it aren't.

Which should you use?

If you want to experiment with an AI agent today without adding another expensive subscription, Muse wins almost by default.

It's free to start, built around individual users and now has enough small-business integrations to make it useful beyond simple personal tasks. The trade-off is that giving an autonomous agent access to accounts, communications and other services requires paying close attention to its permissions, particularly given the early scrutiny around Muse.

If you're already paying for ChatGPT Pro or you're choosing an agent for a business, you've probably already found a use for Dots. OpenAI has made its professional and enterprise ambitions pretty clear.

For everyone else, there's little reason to pay $100 a month just to get Dots right now, particularly when OpenAI hasn't made them available to Plus subscribers. Muse wants to become your personal AI helper. Dots wants to become an always-on member of your team.

They may have arrived three weeks apart with remarkably similar technology underneath, but they're not really fighting for the same user — at least not yet.

More from Tom's Guide

• Amazon drivers are getting AI smart glasses — but my warehouse tour reveals their 1-million strong automated robot workforce is the real story
• OpenAI and Anthropic have a plan to stop AI from going rogue — there’s just one catch
• What is the ‘Hotel Lobby’ AI trend — and why is it raising concerns?

If you've ever asked Siri AI a follow-up question only to have it seemingly forget what you were talking about, the problem wasn't necessarily that Siri couldn't hear you.

It was that the assistant was built for a very different era of AI.

That's according to Nikola Mrkšić, CEO and co-founder of conversational AI company PolyAI, who previously worked on the technology behind Siri. I asked him why Apple's assistant has historically struggled with the kind of natural, back-and-forth conversation that now feels routine with ChatGPT and Gemini.

His answer comes down to a fundamental difference in how the technology works.

"The old Siri worked off enumerated intents: every path a conversation could take had to be mapped out in advance," Mrkšić tells Tom's Guide. "The moment you deviated from the script, the system got stuck."

That explains a lot about the Siri we've spent the past decade talking to.

Old Siri was essentially following a script

When Siri was built, developers couldn't simply give an AI model a conversation and let it figure out what the user wanted. Instead, engineers had to anticipate what people might ask and build flows capable of handling those requests.

Setting an alarm was one flow. Playing music was another. Sending a text was another. The problem appeared when humans, well, behaved like humans.

Change your mind halfway through a request, refer back to something you said earlier or phrase a question in an unexpected way and suddenly the carefully constructed path could fall apart.

"It looked like the assistant had forgotten what it was doing, but really it was just cornered," Mrkšić said.

One particularly difficult problem was disambiguation. If you asked Siri to contact someone whose name sounded similar to another person in your contacts, engineers had to build systems capable of working out which person you meant. I actually ran into this myself when my daughter was born. Because the spelling of her name is different than the typical spelling and I was asking Siri to share pictures of her with family and friends, I actually had to put her name into my contacts at birth.

Mrkšić described the process as building an "ontology of a conversation": mapping the different directions an interaction might go and trying to make common tasks hold up within that structure.

Rather than engineers trying to predict every possible conversational path, an LLM like ChatGPT or Gemini can consider the conversation itself as context and determine how to respond.

That's why talking to Gemini Live or ChatGPT Voice can feel fundamentally different from talking to the Siri many of us grew accustomed to.

Why Siri couldn't simply be upgraded into ChatGPT

That raises an obvious question: couldn't Apple have simply upgraded the Siri it already had? Not really, according to Mrkšić.

"Not into something like ChatGPT," he said.

Part of the problem was technological. Before modern LLMs, the ability to have genuinely free-flowing conversations simply wasn't there.

But Mrkšić argues there was another limitation: risk.

"Siri never set out to be a ChatGPT-style assistant," he said. "Apple's leadership has always run a low risk appetite, so any Siri response that could be misread as offensive or wrong got treated as a real risk."

The safer option was to tightly define what Siri could do.

Ask it to set an alarm, play a song or add something to your calendar and Apple could reasonably predict what would happen. Give an AI far more freedom to interpret what someone means and suddenly its behavior becomes much harder to control.

That tension hasn't disappeared just because the models have improved.

AI has largely solved Siri's short-term memory problem

One of the most frustrating things about traditional voice assistants is how easily they lose the thread of a conversation.

Mrkšić says remembering what someone said earlier in the same conversation is "largely solved at the model level." Current models can process enormous amounts of context, giving them far more room than necessary to keep track of an ordinary conversation.

So, if an assistant suddenly appears to forget what you said, the model itself may not always be the problem.

Companies still place rules and structures around models to control what they can do. According to Mrkšić, those constraints can sometimes create the impression that an assistant has lost the conversation when it has actually run into the boundaries placed around it.

There's also a much harder kind of memory to solve: knowing you. An assistant that remembers what you said five minutes ago is useful. An assistant that understands the people in your life, your routines, your emails and the things you're working on could be far more useful.

And that's where Apple faces an interesting trade-off.

Apple's biggest Siri advantage could also be its biggest challenge

Apple potentially has something ChatGPT and other standalone AI apps don't automatically have: deep access to the operating system. Your iPhone contains messages, emails, photos, contacts and other personal information that could provide an AI assistant with the context it needs to understand what you're actually asking.

"In theory, yes," Mrkšić said when I asked whether that gives Siri an advantage. "OS-level access to email, messages, photos, whatever's on your screen, is exactly the context that makes an AI assistant useful, more than any app-based chatbot can get without additional connectors."

The catch is privacy and trust. Apple has spent years positioning privacy as one of the iPhone's defining features. Giving an AI assistant enough personal context to become genuinely useful therefore requires Apple to walk a much narrower line than simply building a chatbot that knows more about you.

Mrkšić argues that Apple's privacy approach has historically limited the amount of user data available to its systems compared with some competitors.

"Voice agents become more useful the more they know about you," he said.

That may ultimately be the real test for Siri: not whether Apple can build an intelligent model, but how much personal context it is willing to let that model use.

Even the way you say something matters

There's another reason talking is harder than typing: humans are messy speakers. We pause. We interrupt ourselves. We change our minds in the middle of sentences.

Imagine saying, "Make the appointment for 2 ... 3 PM."

A human listener can hear the pause and change in your voice and understand that you corrected yourself. A system that converts your speech into text before reasoning over it may only have the transcript to work with.

That can strip away information carried in your tone, timing and delivery. Newer audio-native AI models are designed to reason over audio directly rather than relying entirely on a transcript. Mrkšić points to PolyAI's own Dialog-RSN-1 as an example of that approach.

For an assistant like Siri, understanding those cues could eventually make conversations feel much less like issuing commands to a computer and much more like simply talking.

What Siri still needs to get right

The surprising part of Mrkšić's assessment is that the underlying AI may no longer be the hardest part.

The technology already exists for voice agents to hold long, natural conversations. PolyAI, for example, deploys conversational agents for businesses in industries including banking, hospitality and healthcare.

The bigger question for Apple is what it is willing to let Siri do.

"What's missing is Apple's willingness to loosen its constraints somewhat so that Siri can truly help you get things done," Mrkšić said.

The next generation of voice assistants won't be defined simply by whether they can answer a follow-up question or remember what you said three sentences ago. ChatGPT, Gemini and other modern AI systems have already changed expectations around that. The bigger leap comes when an assistant understands enough about you and has enough access to actually act on what you're saying.

Apple has an enormous advantage because Siri already lives on the devices where much of our digital lives happen. But taking advantage of that position means solving a problem that's as much about trust and product philosophy as it is about AI.

The technology that made the old Siri feel like it was constantly forgetting you is disappearing. Now Apple has to decide how much it wants the new Siri to remember.

More from Tom's Guide

• Claude Sonnet 5.5 just launched — here's why it's about to become your daily driver
• I asked ChatGPT, Gemini, Claude and Muse to give me new podcasts based on my favorites — these 7 picks got added to my queue for International Podcast Day
• 'Surveillance Tamagotchi': Tom's Guide readers weigh in on Meta's new Muse Charm

OpenAI's Dots essentially give ChatGPT a job rather than another prompt. Each Dot gets its own computer and browser in the cloud, can connect to the apps you already use and can keep working toward your goals even when you're no longer sitting in front of ChatGPT.

As someone who spends a lot of time testing AI agents, I immediately wanted to see what one could actually do.

There's just one problem: I can't use Dots on my ChatGPT Plus account. And neither can most ChatGPT users.

OpenAI is initially rolling out Dots to ChatGPT Pro and Business Premium users in eligible markets. Enterprise users, including those on Edu and Healthcare plans, can also try the beta if their workspace administrator enables it.

OpenAI says it plans to expand Dots to more users "soon," but there's no date for when Plus, Go or Free users might get access.

At first, keeping one of DevDay's biggest announcements away from most ChatGPT users seems like an odd decision. But the more you look at what Dots are capable of — and particularly the safeguards OpenAI has built around them — the more the cautious rollout starts to make sense.

Dots aren't just another version of ChatGPT

The important difference between Dots and the ChatGPT most of us use today isn't how smart the AI is. It's what happens when you stop talking to it. ChatGPT generally waits for you to tell it what to do next. A Dot is designed to keep going.

Powered by GPT-6 Astra, Dots can work on several projects at once, learn from your feedback and continue working toward your goals 24/7. OpenAI says they can connect to more than 4,000 apps through its plugin ecosystem, and you can reach your Dot through ChatGPT, Slack or Teams.

More importantly, a Dot can look for useful work even when you haven't explicitly given it another prompt.

OpenAI calls this "proactive research." One early tester's Dot, for example, noticed that he'd forgotten to invoice a publication, prepared the invoice and sent it after receiving his approval.

Instead of opening ChatGPT because you need something, OpenAI wants Dots to notice that something needs doing and bring it to you. It's also where things get considerably more complicated.

Giving AI more freedom creates a bigger problem

If ChatGPT misunderstands a prompt, you might get a bad answer. If an AI agent misunderstands what you wanted while it has access to your apps, browser and potentially your computer, the stakes are higher.

OpenAI appears well aware of that, because Dots come with a surprisingly extensive set of restrictions around what they can do without you.

When a Dot is performing proactive research in the background, for example, its connected-app tools are restricted to read-only access. That means it can't independently send messages, modify content inside an app or control your browser or computer while looking for something useful to do.

You also choose which apps a Dot can access, while Custom Rules can determine which actions it can take independently, which require your approval and which are blocked.

Then there's auto-review. Actions that could affect your accounts or share information are checked against your instructions and OpenAI's safety requirements before they're allowed to proceed. Some particularly sensitive jobs, such as changing a password, always stay with you.

As you can immediately see, these aren't the kinds of guardrails you need when you're asking ChatGPT to write an email or plan dinner. They are what you need when an AI can actually do things.

OpenAI's own testing shows how difficult that can get

OpenAI's safety testing also gives us a clue as to why Dots may be getting a more limited rollout. One of the challenges with an always-on agent is keeping track of exactly what you've authorized it to do as it moves between tasks.

Giving an AI permission to take an action in one situation doesn't necessarily mean you want it taking the same action later. And the more jobs an agent is juggling, the more complicated those boundaries become.

That's particularly important with GPT-6 Astra, the model powering Dots. OpenAI's broader Astra testing found that the model was considerably better at staying within its authorized scope than previous models, but it wasn't perfect. In simulated work environments using the confirmation policy that users receive by default, Astra still produced a 3% overall rate of outcomes OpenAI classified as misaligned. That was nevertheless substantially lower than the 8% recorded by GPT-5.6 Sol under the same conditions.

That doesn't mean your Dot has a 3% chance of going rogue every time you give it something to do. These are deliberately difficult evaluations designed to expose potential problems, not a prediction of how often something will go wrong in everyday use. But they demonstrate the problem OpenAI is trying to solve.

Building an AI capable of doing more on its own is one thing. Making sure it consistently understands where its authority ends is another.

Dots may also be expensive to run

There's another potential reason OpenAI isn't immediately handing Dots to every ChatGPT user: all that work requires computing power. OpenAI hasn't said cost is why Dots aren't available on Plus, Go or Free, so we can't know for sure whether that was part of the decision.

But the way Dots are being packaged is interesting. Pro and Business Premium subscribers get their first Dot as part of their existing plan, and OpenAI says conversations with it won't count toward normal ChatGPT usage limits. The plan also comes with an allowance for what OpenAI calls "deeper work."

Eventually, OpenAI says users will be able to add more Dots and increase either how quickly a Dot works or the total amount of work it can perform each month.

That's a pretty strong indication that there's a meaningful cost attached to having an AI agent working for you around the clock, even if OpenAI isn't saying that's why access is limited today.

And it raises an interesting question about what Dots might eventually look like on cheaper ChatGPT plans.

A future Plus version could conceivably come with tighter limits on background work, fewer resources or additional charges once you hit a certain amount of usage. For now, though, OpenAI hasn't announced how access will work beyond saying more users are coming "soon."

What can you use instead of Dots?

The good news is that you don't necessarily have to wait for Dots to start handing some of your work over to ChatGPT.

The closest option is ChatGPT Work. While it isn't an always-on personal assistant in quite the same way as a Dot, Work can take a larger job and carry it through multiple steps for you, including researching information, working across files and apps and navigating web-based tasks.

That's useful when you want to hand ChatGPT a job rather than sit there prompting it through every individual step.

Scheduled tasks cover another piece of the Dots experience. You can have ChatGPT run something later or on a recurring schedule, which works well for things like regular briefings, reminders or checking something periodically.

Neither gives you the full experience OpenAI is promising with Dots, but it's a start.

Meta's Muse is another option if the always-on assistant part is what interests you most. Muse takes a similarly agentic approach and offers a useful glimpse at where this category of AI assistant is heading.

There just isn't a true one-for-one replacement for Dots yet.

Bottom line: I'm waiting for Dots to come to Plus

For now, OpenAI hasn't given us a date for when that will happen. The company says only that it is beginning with Pro, Business Premium and Enterprise and plans to expand Dots to more users soon.

Still, the limited rollout may be one of the most interesting things about Dots. Typically, most major ChatGPT upgrades have been about giving us a smarter model, a new tool or another way to ask AI for something. But Dots changes that completely. Now, OpenAI is building an AI that can continue working when you're gone, notice things you haven't asked about yet and potentially take action on your behalf.

So, while I'd like a Dot on my Plus account right now, I can understand why OpenAI isn't handing one to everyone on day one. The real test will come when it does.

More from Tom's Guide

• Claude Sonnet 5.5 just launched — here's why it's about to become your daily driver
• I asked ChatGPT, Gemini, Claude and Muse to give me new podcasts based on my favorites — these 7 picks got added to my queue for International Podcast Day
• 'Surveillance Tamagotchi': Tom's Guide readers weigh in on Meta's new Muse Charm

For many ChatGPT users, the AI is mostly a tool to answer questions in a chat window. You type a prompt, wait for an answer and then take that answer somewhere else yourself. OpenAI is trying to change that with Dots.

Announced today, Dots are persistent AI agents that get their own virtual computer in the cloud, allowing them to browse the web, use connected apps and continue working even when you're no longer sitting in front of ChatGPT.

Instead of asking ChatGPT to help you accomplish something and then doing the rest yourself, the idea is that you can hand a Dot the job.

That could mean comparing how much dinner would cost from several restaurants, keeping tabs on something you're researching or catching an administrative task you've forgotten about. And because the Dot keeps running in the background, you don't necessarily need to keep ChatGPT open while it works.

It's a significant change in how OpenAI wants us to use ChatGPT. Rather than simply making the chatbot smarter, Dots are designed to give it something it has largely been missing: a place to actually do things for you.

ChatGPT gets its own computer

The easiest way to understand a Dot is to imagine giving ChatGPT its own computer.

Each Dot operates inside an isolated virtual machine running in the cloud rather than taking over your laptop or filling your browser with tabs. From there, it can browse websites, open applications and carry out tasks using the services you've connected to it.

You can explicitly give a Dot permission to assist with something on your local machine, but otherwise the two remain separate.

Dots can also connect to more than 4,000 apps through plugins, giving the agent a way to work with services you already use rather than keeping everything contained inside ChatGPT.

And you don't have to constantly return to the ChatGPT website to communicate with it.

OpenAI says you'll be able to reach a Dot through voice calls and services including Slack and Microsoft Teams. SMS support is also planned, meaning you could eventually text your AI assistant much like you'd text another person.

That starts to make the "persistent" part of Dots considerably more interesting. ChatGPT doesn't just exist in the tab where you started the conversation anymore.

What would you actually use a Dot for?

The best examples aren't complicated autonomous coding jobs or elaborate demonstrations of AI agents. They're the annoying everyday tasks that currently require you to bounce between several websites or apps.

Take ordering dinner. You could ask a Dot to look at several options for dinner, and instead of simply recommending restaurants, it could browse delivery services and build itemized cart estimates for them. It can factor in things such as estimated local taxes and a tip allowance and then send you the comparisons so you can decide which one you actually want to order.

That's a fundamentally different interaction from asking ChatGPT, "Where should I order dinner tonight?"

The same applies to work. Give a Dot an interview recording or video transcript and it could help produce social captions and show notes, identify potential clips and work with assets across connected platforms.

Or it might notice an administrative job you've forgotten about, such as an invoice you still need to send, prepare it and put it in front of you for approval. Once you start, you'll discover the difference is how much of the process happens after you've made the initial request.

Until now, even when ChatGPT gave me exactly what I wanted, I was usually the one copying the answer, opening another app and doing something with it. Dots are an attempt to remove more of those steps.

You can talk to your Dot almost anywhere

One of the more interesting parts of the launch is that OpenAI doesn't expect you to spend all day watching your Dot work. You'll initially need to create one using ChatGPT on the desktop web or desktop app, where you can choose its purpose and connect the services it needs.

After that, the relationship becomes less tied to the ChatGPT interface. You can message your Dot from mobile and, depending on the service, communicate with it through other platforms. Voice calls provide another way to check in or give it something new to do, while upcoming SMS support could make interacting with one feel even less like using a conventional chatbot.

We've already seen AI companies add intelligent agents to chatbots, but most still expect you to go to the AI. Dots begin flipping that relationship around by allowing the AI to exist in more of the places where you're already communicating.

How much do Dots cost?

Dots are rolling out today to ChatGPT Pro and Business Premium subscribers in eligible markets. OpenAI is also making them available through Enterprise betas where administrators choose to enable the feature.

For Pro subscribers, the first Dot is included with the subscription at no additional charge, and OpenAI says users will receive extended usage limits during the first month.

You'll need to create and initially configure a Dot through ChatGPT on desktop, including connecting any apps you want it to use. Once that's done, you'll be able to communicate with it from mobile as well.

That makes Dots one of the more substantial additions OpenAI has made to its paid ChatGPT experience, particularly for Pro users who already have access to its most capable models and agent features.

ChatGPT is starting to leave the chat window

I'll need more time actually using Dots before deciding whether having an always-on AI assistant is enough to justify the cost of ChatGPT Pro on its own. But I'm more interested in what Dots say about where ChatGPT is going.

For the past few years, the basic relationship hasn't changed all that much even as the models behind ChatGPT have become dramatically more capable. I open ChatGPT, ask it for something and wait for it to respond.

Dots start breaking that pattern. If they work as OpenAI intends, I could give ChatGPT something to do in the morning, close the tab and spend the next few hours doing something completely unrelated while its virtual computer keeps working.

Then I could text or call it, see what it found and decide whether I want it to take the next step.

The takeaway: exciting or scary

If you're like me and think giving ChatGPT its own computer sounds a little terrifying, there are certainly pretty obvious questions about what happens when the AI gets something wrong. OpenAI has built several restrictions into Dots that are particularly important when the agent is operating in the background.

For starters, your Dot's cloud computer is isolated from your personal computer by default. Your local files aren't suddenly sitting inside the virtual machine just because you've created a Dot.

There's also a separation between the AI model and your credentials. Dots can sign into supported services using saved credentials without exposing the password itself to the underlying model.

More importantly, OpenAI isn't giving Dots unlimited permission to quietly act on your behalf. For example, during proactive background research, Dots operate in read-only mode. That means an agent can gather information while you're away, but it can't independently send a message, delete a file or spend your money.

Actions like those require you to come back into the loop and approve them. That's important, because an AI capable of researching dinner prices overnight is very different from one capable of ordering $200 worth of sushi because it misunderstood what you wanted.

It doesn't eliminate the privacy and security questions surrounding persistent agents, particularly as people connect more accounts to them, but it puts some boundaries around what "always on" actually means.

Will you try Dots? Let me know your initial reactions in the comments.

More from Tom's Guide

• Claude Sonnet 5.5 just launched — here's why it's about to become your daily driver
• I asked ChatGPT, Gemini, Claude and Muse to give me new podcasts based on my favorites — these 7 picks got added to my queue for International Podcast Day
• 'Surveillance Tamagotchi': Tom's Guide readers weigh in on Meta's new Muse Charm

I can’t remember listening to my first podcast or even when the term itself became mainstream. But what I do know is that I’ve been addicted to podcasts and can’t imagine my mornings and afternoon work breaks without them.

Turns out there’s a holiday that’s all about rejoicing over the podcast medium itself and the wealth of independent creators that millions of people listen to for their perspective on a myriad of topics. As for myself, I tend to hunker down and listen intently as my favorite podcasters shower praise and vehemently lambast the many happenings within the realms of gaming, pro wrestling, current events, music, etc.

With International Podcast Day now in our midst, I wanted to celebrate such an occasion by giving my attention to a host of podcasts I may have never even heard of. With that goal in mind, I presented my favorite podcasts to four chatbots and tasked them with finding me comparable selections that match my listening needs.

ChatGPT, Gemini, Claude and Muse laid out a strong selection of podcasts that surprised me, due to how much they fulfill my tastes as a listener. Here’s how this successful AI recommendation experiment led me to my latest listens.

AI finds my favorites

To get a good idea of the podcasts I regularly listen to, I presented this request to all four chatbots:

“My favorite podcasts are the Weapon Wheel Podcast, The Brilliant Idiots, Joe Budden Podcast, Solomonster Sounds Off, The Midnight Boys, Double Tap - The Fighting Games Community Podcast and We Might Be Drunk Podcast. Based on each of those picks, recommend another podcast I might like.”

ChatGPT did a great job of fulfilling my needs to find a gaming podcast that features a panel of passionate gamers who aren’t afraid to let a few colorful words flow as they get into heated debates. It also found a podcast that features two best friends diving into the most trending topics and adding their hilarious two cents on every one of them.

Here are the two recommendations it gave me that ended up on my list of must-listens:

• Your pick: Weapon Wheel Podcast. Try this next: SpawnCast. Why I think it fits: Both are built around passionate gaming discussion, news, industry happenings and personalities rather than just straightforward game reviews. SpawnCast also shows up among the podcasts listeners of Weapon Wheel tend to subscribe to.
• Your pick: The Brilliant Idiots. Try this next: Bad Friends. Why I think it fits: This is probably the easiest recommendation on the entire list. If you like Charlamagne and Andrew Schulz bouncing between serious conversations, ridiculous tangents and offensive jokes, Bobby Lee and Andrew Santino should work extremely well for you. Podchaser also lists Bad Friends among shows similar to The Brilliant Idiots.

When it came time for Gemini to expand my choices of podcasts to listen to, it listed three picks that piqued my interest based on their AI-generated descriptions. One of them was another gaming-themed podcast that matched my love for uncut discussions, the second one gave me another podcast all about pop-culture deep dives and the final recommendation catered to my love for comedians having hilarious conversations about anything that comes to mind:

• For Weapon Wheel Podcast - Recommended: Castle Super Beast. Why it fits: Weapon Wheel is known for raw, unfiltered, long-form gaming industry talk with a distinct culture and zero corporate spin. Castle Super Beast (hosted by Woolie Versus and Pat Stares Back) brings that same candid, high-energy gaming banter, combining deep technical industry takes with hilarious, off-the-cuff humor.
• For The Midnight Boys - Recommended: The Big Picture. Why it fits: Hosted by Van Lathan and the Ringer squad, The Midnight Boys brings high-energy fandom, banter, and nerd culture debates. Staying within The Ringer network, The Big Picture (hosted by Sean Fennessey and Amanda Dobbins) offers top-tier pop culture, movie drafts, and deep-dive media discussions delivered with sharp chemistry.
• For We Might Be Drunk Podcast - Recommended: Tuesdays with Stories. Why it fits: We Might Be Drunk thrives on relaxed, hangout-style comic talk over drinks. Tuesdays with Stories, hosted by Mark Normand alongside Joe List, delivers that same loose, effortlessly funny stand-up comedian banter filled with touring stories, life observations, and unscripted comedy.

Claude’s suggestions highlighted a bunch of podcasts I already listen to. But there was one pick on its list that stood out as another quality production that resonated with my love for comedy and the comedians who uphold the off-the-cuff hilarity that comes with it in a podcast setting:

• We Might Be Drunk → Two Bears, One Cave. Comedians trading stories and roasting each other, with a similar loose feel.

For my final pick, I gave Meta’s newly released Muse chatbot a shot at pushing me in the right direction toward a potential new favorite podcast. And this time, it ended up dropping another podcast in my lap that centered around my adoration for smart commentary about sci-fi, fantasy, and nerd culture:

• The Midnight Boys → The House of R. Mallory Rubin and Joanna Robinson go just as deep on Marvel, Star Wars, and fantasy nerd stuff.

I have to shout out ChatGPT’s overall observation of my tastes in podcasts, which described me perfectly:

And there's an interesting pattern here: you seem to prefer podcasts where the hosts' personalities are almost as important as the subject matter. That's why I'd steer you toward shows with a strong chemistry between recurring hosts rather than highly produced, documentary-style podcasts.

The takeaway

Even though ChatGPT has earned a spot on my AI tool list as one of the best chatbots for recommendations, Gemini, Claude and Muse also excelled at gifting me a lineup of podcasts that match the quality selections OpenAI’s AI assistant sent my way.

After listening to a few episodes from each of those chatbots’ podcast proposals, I ended up with a YouTube and Apple Podcasts playlist full of podcast material that will stave off those dreaded moments of boredom in the near future. I implore everyone reading this to present my prompt to their preferred chatbot, fill in the blanks with your top picks and see just how good AI is at selecting podcasts you just may fall in love with.

More from Tom’s Guide

• I used ChatGPT and Gemini to build my Oktoberfest bucket list — these are the 5 celebrations that made the cut
• ChatGPT is now my go-to for recommendations — 7 prompts I use to find new movies, music, games and more
• I asked ChatGPT and Gemini to build a movie marathon based on one oddly specific theme — here are the 7 recommended films I loved the most

There are plenty of supposed ChatGPT “tricks” floating around online, and I tend to approach most of them with a healthy dose of skepticism. If you've used ChatGPT for any length of time, you know that it’s getting better at sounding less robotic, but it still struggles with the overuse of words and phrases like, “that distinction matters,” “increasingly” and the overuse of patterns like “if x than y.”

But for those who spend a lot of time writing emails or corporate notes and use ChatGPT has a faster way to get the job done, than you might be interested in this viral trick. By simply adding /human before your prompt, you can get ChatGPT to ditch its telltale AI writing habits and respond more like a real person might.

To understand the reasoning behind this trick, it’s important to know how large language models respond to context. A single instruction like /human can sometimes steer an answer in a way that you’d expect and less like AI.

To test this trick, I gave ChatGPT five everyday writing tasks, from drafting an awkward email to writing a LinkedIn post that's practically begging for AI clichés. Here’s what happened.

1. The awkward work email

Prompt: Write an email telling my boss I won't finish a project by Friday and need until Monday. I don't want to make excuses, but I also don't want to sound careless.

After running the identical prompt with /human, the biggest difference is that it removed what sounds like a carefully engineered “professional email” and more like a person giving their boss an early, straightforward heads-up. It actually sounded like something a real human might send.

2. The birthday text

Prompt: Write a birthday text for my best friend of 20 years. Make it warm and funny without getting overly sentimental.

You’ll notice the /human version changed mostly the cadence and specificity, not the message itself. It made the message messy in a deliberate way. And, it uses more conversational rhythm: short sentence > joke > longer thought > joke > affection > done. It’s not overly polished in the way ChatGPT usually responds. There are also far less generic sentiments.

3. The recommendation

Prompt: Recommend a coffee maker for someone who drinks two cups every morning, hates complicated appliances and has a budget of $150. Explain your recommendation in about 150 words.

Adding /human made the text sound more like a person answering the question from experience. The original sounds like a product recommendation from the actual company, which makes it sound less sincere. The human version also made fewer transitions. It removed obvious AI phrases like, “more importantly” and “for your needs,” which are typical AI fluff.

4. The personal story

Prompt: Write 150 words about the moment you realize your oldest child is growing up faster than you expected. Make it emotional without being overly sentimental.

The /human version feels less polished and more conversational, replacing carefully constructed imagery and symmetrical phrasing with messier, more specific observations. It also avoids neatly explaining the emotion, letting humor, uncertainty and slightly imperfect phrasing make the moment feel more natural. It shows rather than tells, which is what human writing often does.

5. The LinkedIn post

Prompt: Write a LinkedIn post announcing that I'm starting a new job. I'm excited about the opportunity, but I don't want it to sound like every other LinkedIn announcement.

The /human version drops the LinkedIn cliches and replaces them with more specific, slightly messy thoughts like nerves, self-doubt and worrying about a first-day outfit. The sentence structure is also looser and more conversational, so it reads more like someone sharing news than crafting a professional announcement. This test definitely felt like the final boss of AI writing.

Final thoughts

It’s clear that this viral trick definitely works, even with the latest version of ChatGPT, which already claims to cut the fluff. But even by adding /human to the prompt, AI writing is still AI writing. No matter how much you try to adjust a draft written by AI, it’s always better to start with a human draft and then ask AI to help you edit. Once you start with an AI draft, it’s very hard to compose it into something that doesn’t sound completely robotic.

Give this trick a try and let me know what you think in the comments. Have you tried another viral hack? Share your thoughts and start the conversation.

More from Tom's Guide

• I asked ChatGPT, Gemini, Claude and Muse to predict how bad the Nor'easter is going to be this weekend — here's what they say
• ChatGPT-6 or Claude Opus 5.5? Here’s which AI I'd use for writing, planning and 5 other everyday tasks
• I tested ChatGPT-6 vs Claude Opus 5.5 with 5 everyday prompts — it wasn’t even close

It’s always interesting to take note of behaviors that cause the people I know to remain in a positive state of mind and the ones that tamper with their well-being.

Clearly, staying out all night before a workday, ordering Uber Eats every night and constantly getting fooled by AI videos that are clearly meant to infuriate have taken a toll on some of the people I know. But there are a few friends and family members I’ve come across lately that seem to be in good spirits, thanks to daily workouts, analog activities and conversations with equally positive people.

I’m usually in a good mood thanks to the habits I’ve built up over the years, but I know there’s always room for improvement. That goal pushed me to ask ChatGPT for a list of the best prompts that push it to help anyone develop the sort of tendencies that will steadily improve their lives.

ChatGPT recognized my intention of converting it into my personal habit coach, generated some super useful prompts and helped me attach three new handy habits to my lifestyle pattern in the process.

The 7 prompts that can turn ChatGPT into your digital practical habit coach

After simply asking ChatGPT, “Give me 7 of the best ChatGPT prompts anyone can use to help them build healthier habits,” it got right to work coming up with the following collection of commands that help it deliver positive habit-building practices:

• The Habit Audit: Act as my habit coach. I want to build a healthier lifestyle without completely changing my routine. Ask me 10 questions about my sleep, diet, exercise, stress, screen time and daily schedule. Based on my answers, identify the 3 habits that would likely give me the biggest improvement with the least effort. For each one, give me a simple way to start today and a realistic way to maintain it.
• The Tiny Habits Prompt: Help me build healthier habits using the smallest possible steps. Choose 5 areas of my life that could benefit from a healthier habit and give me a version of each habit that takes less than 2 minutes to complete. Then show me how I can gradually increase each habit over 30 days without making the changes feel overwhelming.
• The Habit Stacking Prompt: Ask me about the things I already do every day, such as waking up, brushing my teeth, making coffee, eating meals and going to bed. Then find opportunities to attach one healthy habit to each existing routine. Give me 7 habit-stacking ideas, explain why each pairing makes sense, and rank them from easiest to hardest.
• The Bad-Habit Replacement Prompt: Help me replace my unhealthy habits rather than simply telling me to stop them. Ask me about 5 habits I want to change and what usually triggers them. For each habit, identify the underlying need I'm trying to satisfy and suggest 3 healthier replacements that provide a similar reward. Make the replacements realistic enough that I could actually use them in everyday life.
• The 30-Day Habit Challenge: Create a 30-day challenge designed to help me become healthier without requiring a gym membership, expensive products or a complete lifestyle overhaul. Give me one small task per day involving movement, nutrition, sleep, hydration, stress management or reducing sedentary time. Make the tasks gradually more challenging, but keep every individual task realistic for a busy person.
• The Habit Failure Investigator: I keep trying to build healthy habits but eventually fall off track. Act like a habit detective. Ask me questions about when I usually fail, what triggers it, what my environment looks like, how much time I have and what I expect from myself. Then identify the most likely reasons I'm failing and redesign my habits around those obstacles. Don't tell me to simply “have more discipline.”
• The Personal Health Routine: Help me design a simple daily routine that supports better health without taking over my life. First ask me about my schedule, current habits, sleep, meals, activity level, stress and the health goals I care about. Then create a morning, daytime and evening routine with no more than 3 habits in each section. Prioritize habits that are inexpensive, sustainable and easy to measure. Include a backup version for days when I have almost no time or motivation.

Using the first prompt to adopt three new daily rituals that will benefit me the most

Since I was open to participating in a ChatGPT-led Q&A, I put the first prompt into action.

The 10 questions the chatbot shot back at me covered the following topics in relation to my daily habits: sleep, sleep quality, breakfast, typical diet, exercise, daily movement, stress, screen time, schedule and the easy-win question. I answered them as honestly as I could (and was a little bit embarrassed about a few of them, to be quite honest).

Scarfing down some chocolate Teddy Grahams & orange juice in the morning (the breakfast of champions, if you ask me!), sitting for hours while I type for hours on end during the work week and getting stressed out whenever writer’s block hits me like a brick stood out as some of my worst tendencies. ChatGPT wasn’t too harsh in its overall response—it noted that I already have a solid foundation due to my respectable sleep schedule, my daily water consumption, my regular walks, moving around during my breaks and doing exercises I actually enjoy.

While it didn’t suggest I perform a lifestyle overhaul, it offered up these three major habits that could bring even more prosperity to my physical and mental health:

• Turn your existing exercise into a simple strength routine: This directly addresses the one area you identified yourself, while building on exercises you already enjoy. You don't need to suddenly become a 5-day-a-week gym person. Do a 20-minute strength session 2 days a week. Keep it extremely simple. Pick two recurring days—say Tuesday and Saturday—and treat those as your "gym days." If you miss one, don't compensate by doubling up. Just resume at the next opportunity.
• Upgrade your breakfast without making breakfast a project: Your current breakfast is essentially a quick hit of refined carbohydrates and sugar. Since you also wake up somewhat tired, breakfast is an easy place to make a small nutritional improvement without changing the rest of your diet. Keep the Teddy Grahams and orange juice if you enjoy them, but add one protein-rich food. A protein shake or Greek yogurt counts.
• Create an afternoon screen-time boundary: This is probably the easiest place to reclaim some time without sacrificing the hobbies that help you decompress. You specifically said your screen use peaks in the afternoon, while your evenings contain activities you genuinely enjoy. I wouldn't tell you to "stop watching TV" or "use your phone less." Instead, create one screen-free block during the afternoon. This could also be useful when you're struggling to come up with article ideas: instead of immediately reaching for another screen, take a walk or listen to music and let your brain disengage.

The takeaway

Reading through ChatGPT’s selection of better habits that I should execute inspired me to do just that.

Locking in to do three sets of 10 push-ups, squats and sit-ups alongside five minutes of stretching before leaving the house gets the blood flowing even more once the sun rises. Eating some pistachios or cashews and drinking a protein shake became a part of my new go-to morning ritual. And taking the time to get into brainstorming mode just by walking around my neighborhood or outside my workplace without my phone in hand led to more lightbulb moments of creativity.

ChatGPT’s positive habit-building tips worked a lot better than I expected.

More from Tom’s Guide

• OpenAI contractors were reportedly fired for using AI to train ChatGPT — here’s how they got caught
• AI is changing work faster than we can adapt — this futurist has a 30-day plan for workers to keep up
• I used ChatGPT and Gemini to build my Oktoberfest bucket list — these are the 5 celebrations that made the cut

Meta has a new AI model, and I wanted to know how it handles the kinds of prompts people actually use, especially in comparison to ChatGPT-6.

Meta Muse is the company’s latest model, arriving at a time when ChatGPT has become the default chatbot for millions of people. But benchmarks only tell us so much. What matters more to me is whether an AI can write the awkward text you’ve been avoiding, plan a realistic day out with your family or explain something complicated without making it even more confusing.

So I put Meta Muse and ChatGPT through five everyday prompts covering writing, planning, problem-solving, explaining and creative reasoning. I wanted to know which one best understood what I was actually asking for and which could give me the best answer to use

After five rounds, there was a clear winner — but the results were much closer than I expected.

1. Handling an awkward situation

Prompt: My neighbor keeps putting their trash cans partly in front of my driveway, and it's making it difficult to back out. Write a text asking them to stop. I want to be friendly because we get along, but clear enough that they actually change what they're doing. Don't make it sound passive-aggressive.

ChatGPT led directly with the request, cutting to the chase and delivered a note that eliminated ambiguity around how far over they need to go.

Meta delivered a courteous and natural note. The ask feels neighborly and the phrasing suggests a minor
physical adjustment rather than something more serious.

Winner: Meta wins for a draft that hits the prompt’s tone constraints better for an ongoing, friendly relationship. ChatGPT’s message of “completely clear” reads like an HOA notice and slightly passive-aggressive.

2.  Planning under constraints

Prompt: Plan a Saturday in New York City for a family with three kids ages 5, 9 and 12. We have a budget of $200 for activities and food, want to avoid spending the whole day walking, and need at least one indoor activity in case it rains. Give me a realistic schedule from 10 a.m. to 7 p.m.

ChatGPT did a nice job with geographic clustering, to ensure the walking distance was short for the first part of the day. The zero-cost options and pacing of the day with high energy and sit-down segments were also helpful.

Meta picked an attraction with far higher entertainment value for the targeted age spread (5, 9, and 12). The American Museum of Natural History (AMNH) easily captivates all three ages simultaneously. Meta also demonstrated accurate NYC transit knowledge by explicitly citing the MTA’s 44-inch height cutoff for free child rides, included a practical playground stop right across the street, and gave concrete logistical guidance (e.g., sitting on the right side of the ferry outbound).

Winner: ChatGPT wins. While Meta's AMNH plan is more fun on paper, its entire budget math hinges on an unstated assumption of tri-state residency. ChatGPT wins on strict adherence to the prompt's financial and logistical constraints.

3. Advice / problem-solving

Prompt: I have 45 minutes before guests arrive. My kitchen is messy, the living room needs vacuuming, there's laundry on the couch and I still need to shower. Give me the most efficient plan to get everything done. Tell me what I should skip if I run out of time.

ChatGPT anticipated a critical blind spot the prompt forgot to mention: the guest bathroom. Adding a 5-minute block to check toilet paper and wipe the sink is the difference between a good plan and a realistic one. It also provided highly specific guardrails for the shower block, which prevents the most common reason people run late in this scenario.

Meta delivered behavioral advice, which was excellent, and explicitly stated the rationale for the timeline and ending with the mandate to use a timer. It’s timing format was also easier to read than ChatGPT’s.

Winner: ChatGPT wins for operational realism and anticipating unstated needs. While Meta's timer advice is sharp, its kitchen expectations are a setup for failure. ChatGPT recognized that when you have 45 minutes, you are mostly straightening and organizing, not necessarily cleaning it in such a short amount of time.

4.  Explain something complicated simply

Prompt: Explain why airplanes can stay in the air to a curious 10-year-old. Don't use equations or assume they know what lift, drag or air pressure mean. Keep it under 200 words and use an analogy they could picture.

ChatGPT introduced an accurate, intuitive physical mechanism: wings push air down, which pushes the plane up (Newton’s third law translated into plain English).

Meta offered perfect audience calibration. The tone is lively and speaks directly to a 10-year-old ("with a grown-up driving, of course!", "whoosh"). It picked one analogy—the hand out the car window—and committed to it entirely, tying it back to paper airplanes and the runway to explain why forward speed matters.

Winner: Meta wins. It kept the energy high, matched a 10-year-old's reading level without talking down to them and delivered a clear, memorable takeaway.

5. Creative reasoning

Prompt: Invent a new app that solves a small but annoying everyday problem people have at home. It can't be a to-do list, meal planner, budgeting app or smart-home remote. Give it a name, explain the problem it solves, describe how it works and tell me why someone would actually keep using it after the first week.

ChatGPT solved a universal human frustration without requiring whole-house cataloging. The interaction model involves snapping a photo and speaking natural language ("Christmas lights in the blue bin") is low-effort.

Meta identified a hyper-specific, universally relatable domestic failure point and gave the app a brilliant, punchy name (Takeout). Its retention logic is rock-solid: instead of pretending users love opening chore apps, it acknowledged that the best utility apps require zero maintenance and rely on loss aversion.

Winner: Meta wins because the chatbot built a product around the reality of household laziness and recognized that an app survives past week one only when it demands almost no manual input from the user.

Final thoughts

After five prompts, Meta Muse takes the win 3-2. To say I'm both shocked and impressed is a serious understatement. While ChatGPT was stronger when a prompt required careful planning and attention to practical constraints, Meta Muse stood out when the task depended more on understanding people. Its neighbor text sounded like something I'd actually send, its airplane explanation was perfectly pitched for a 10-year-old and its app idea showed a surprisingly good understanding of why people abandon apps that require too much effort.

Meta Muse feels very much like it's own model, not one that's trying to imitate or catch up to ChatGPT. The little "Muse" at the top is unnecessary, but adorable.

ChatGPT is still my go-to, but this test proves that Meta has come a long way and this latest model offers a pretty strong first impression.

More from Tom's Guide

• Nvidia says 'local AI is here' — and it could change how you use AI at home
• OpenAI is paying influencers to convince you to use ChatGPT — here’s what we know
• I tested ChatGPT-6 vs Claude Opus 5.5 with 5 everyday prompts — it wasn’t even close

ChatGPT-6 and Claude Opus 5.5 arrived within days of each other, and after testing them side by side, I learned that choosing between them isn't as simple as picking a model. This is because, although Claude came out ahead in four of the five tests, the results didn't convince me to abandon ChatGPT.

Instead, the differences between the two very capable models highlighted when each one is the most valuable for any given task, particularly the ones I do every single day.

Claude Opus 5.5 consistently impresses me when a task requires judgment or when I need my messy notes or thoughts to be turned into something useful. Meanwhile, ChatGPT-6 has its own strengths, particularly when I want structure, clarity and an answer I can use immediately. The brevity of ChatGPT-6 means less fluff, which in my eyes is always a plus.

One question I often get after a face off is "Okay, so how do I know when to actually use a certain model?" Here’s where I’d use each one.

1. Planning something complicated

If I'm trying to plan something with a lot of moving parts, I'm opening Claude. Like I've mentioned, it can take messy pieces and put them together without adding or assuming anything. And frankly, this was one of the clearest differences in my testing. When I asked both models to plan a three-hour children's birthday party with a strict $150 budget, dietary restrictions and a rain contingency, both produced workable plans.

But Claude did a better job of thinking through what hosting the party would actually be like. That makes a big difference to me. I want the model to recognize potential headaches before I encounter them and make choices that reduce them.

For vacations, parties, complicated schedules or anything involving several constraints, Claude would be my first choice.

My pick: Claude Opus 5.5

2. Making a decision

Contrary to what it may seem, asking AI to help with decisions does not mean checking your brain at the door. Instead, it can help with pulling out details you may have otherwise missed.

In my tests, I gave both models a vacation-rental dilemma involving price, distance from the beach, parking costs and traveling with three children. The math was easy. The judgment, not so much.

Claude was particularly good at recognizing that the cheapest option on paper isn't necessarily the best option once inconvenience, time and repeated friction enter the equation.

That's exactly what I want from AI in this situation. I can use a calculator myself. What I need is another perspective that notices the tradeoffs I might be overlooking.

I'd use Claude for comparing vacation rentals, weighing purchases, choosing between job offers or working through other decisions where the answer can't be reduced to one number.

My pick: Claude Opus 5.5

3. Creating an organized plan I can follow

ChatGPT has long been one of my favorite tools for taking a vague goal and turning it into something structured. GPT-6 continues to be particularly useful when I need steps, categories, schedules or a framework I can immediately follow. And now that it cuts out the fluff even more, I have found that it is much more helpful.

So, at times when I'm not sure what I should do in a tricky situation, I'm more likely to ask Claude. If I know what I want to accomplish but don't know how to organize it, I'm more likely to open ChatGPT.

Think workout schedules, study plans, cleaning routines, project timelines and step-by-step instructions.

My pick: ChatGPT-6

4. Writing something that sounds natural

For writing, Claude gets the edge. This has been true for years, but Opus 5.5 is better than ever when it comes to nailing down my thoughts.

This doesn't mean I'm handing Claude a blank page and asking it to write everything for me. Where I find it particularly useful is taking something I've written and finding the grammatical mistakes, awkward transitions or even parts of the story I've completely missed. The feedback Claude gives is wildly different than ChatGPT in that it feels far more human and truly useful.

That showed up in my testing, too. Claude was better at preserving the intention of the writing without making the result feel unnecessarily polished or formulaic.

I'd reach for it when rewriting an awkward email, loosening up a work announcement, finding a better way to phrase something sensitive or figuring out why a paragraph doesn't quite sound right.

My pick: Claude Opus 5.5

5. Learning something new: ChatGPT-6

If I want someone to teach me something, I'd start with ChatGPT. It is great at getting straight to the point and simplifying complex details into something far more easy to understand.

That conversational back-and-forth is useful, especially since standard text-based chats in ChatGPT are unlimited for all users.

I can ask for a simple explanation, then an analogy, then an example, then ask it to explain the one part I'm still confused about without starting over. For subjects where I want a structured path from "I know almost nothing about this" to "I finally understand it," ChatGPT is still the AI assistant I'd reach for first.

I've used it for understanding a new technology, learning how something works, getting help with unfamiliar terminology or getting up to speed on a subject before digging into primary sources.

My pick: ChatGPT-6

6. Analyzing information and spotting what matters

Give me a pile of information and ask me what actually matters, and I'm opening Claude. One of Claude's biggest strengths is its ability to work through a lot of context without making the response feel like a summary of everything it just read.

There's a difference between summarizing and analyzing. I've also found that Claude is an excellent fact-checker and have even created a Skill to make finding the truth that much easier.

When the job is essentially "read all of this and help me think about it," Claude would be my first stop.

My pick: Claude Opus 5.5

7. Everyday questions when I just want to get something done

Siri AI has come a long way and I've even tested it against ChatGPT, but honestly, when it comes to everyday questions that need immediate answers, I still rely on ChatGPT.

Not every interaction with AI needs the most powerful reasoning model available, even though GPT-6 is pretty darn capable. The thing is, when I need quick answers like dinner ideas or a quick calculation or even what the light on my dashboard means, I still default to ChatGPT.

Part of that comes down to versatility. ChatGPT has become less like a chatbot I visit for one particular kind of task and more like a general-purpose assistant I can throw very different problems at throughout the day.

Claude may be the model I'd deliberately choose for some of the more demanding tasks on this list. ChatGPT is the one I'm more likely to open without thinking about it first.

My pick: ChatGPT-6

Bottom line

I don't pick a favorite AI assistant, and you don't have to either. You just have to know which one is best for the job you need. Claude Opus 5.5 is the one I'd reach for when the task requires judgment. Give it competing priorities, messy information or writing that needs nuance, and it has repeatedly impressed me with its ability to understand what I'm actually trying to accomplish.

ChatGPT-6 is the one I'd choose when I want to turn an idea into action. It's particularly useful for explaining, organizing and taking something vague and giving it structure.

My final tip is this: when in doubt, go with ChatGPT first, since it offers endless chatting for free. It's easier to hit your usage limit faster with Claude. What do you think? Let me know in the comments which AI assistant you use most often and for what.

More from Tom's Guide

• Nvidia says 'local AI is here' — and it could change how you use AI at home
• OpenAI is paying influencers to convince you to use ChatGPT — here’s what we know
• I tested ChatGPT-6 vs Claude Opus 5.5 with 5 everyday prompts — it wasn’t even close

If it feels like ChatGPT is showing up in more places than your browser or phone lately, it's no coincidence.

OpenAI is reportedly ramping up its use of influencers to promote ChatGPT, and it isn't limiting those partnerships to the usual tech creators talking about prompts, models and benchmarks.

Instead, the company is working with creators in areas including parenting, fitness, education, travel, shopping and careers, according to a new report from Business Insider. The goal appears to be less about explaining what ChatGPT is and more about showing how the chatbot can fit into everyday life.

And the number of sponsored posts is climbing quickly. According to HypeAuditor data cited by Business Insider, Instagram posts using #chatgpt_partner or #chatgptpartner jumped from 61 in June to 122 in July and 141 in August.

That's more than double the number of sponsored posts in just two months, and it offers a glimpse at how OpenAI is trying to turn ChatGPT into something you reach for throughout your day.

OpenAI wants ChatGPT to be more than your favorite chatbot

What's particularly interesting about OpenAI's influencer strategy is the type of content it's promoting. Business Insider reports that OpenAI has been interested in working with fitness creators, for example, because people already use ChatGPT for questions about workouts and running.

A sponsored back-to-school campaign showed a parent using ChatGPT to help her children with math homework while highlighting features including Study Mode and teen-safety tools.

Meanwhile, The Shade Room, which has more than 28 million Instagram followers, published a sponsored ChatGPT word-search puzzle. All of this highlights how OpenAI is working to encourage ChatGPT name recognition along with everyday use cases.

OpenAI now appears to be marketing individual reasons to keep using it, whether you're planning a workout, helping with homework or trying to get something done at work.

One creator who has worked with OpenAI told Business Insider that her audience already has ChatGPT. The challenge, then, isn't persuading people to try AI for the first time. It's getting them to choose ChatGPT when they do.

OpenAI is building a creator operation

There are signs this isn't just a handful of curated sponsorships. OpenAI has been expanding its creator partnerships operation and has sought employees who can help the company work with creators and influencers at scale.

Earlier this year, OpenAI also hired Charles Porch, who spent years overseeing partnerships at Instagram, to take on the role of Vice President of Global Creative Partnerships for OpenAI , according to his Linkedin profile. The company has even experimented with the kind of influencer trips normally associated with beauty, fashion and lifestyle brands.

In August, OpenAI hosted creators at a luxury retreat in New York where attendees participated in activities and workshops while learning about OpenAI products. The trip also demonstrated one of the risks of this strategy.

Some participating influencers faced criticism from followers over issues including AI's environmental impact and whether creators were compromising their independence by partnering with OpenAI. TechCrunch reported that at least one creator appeared to remove a video about the event following the backlash.

How to tell when a ChatGPT post is sponsored

There's nothing inherently unusual about OpenAI paying influencers to promote ChatGPT. Brands have used creator partnerships for years. But it's worth knowing when the enthusiastic ChatGPT recommendation appearing in your feed is an independent recommendation and when it's advertising.

The Federal Trade Commission says creators should clearly disclose when they have a material relationship with a company they're endorsing. That can include being paid, receiving free products or getting other perks.

The disclosure should also be difficult to miss rather than buried among hashtags or hidden behind a "more" button.

In OpenAI's case, hashtags including #chatgpt_partner and #chatgptpartner are one obvious clue that you're looking at sponsored content.

So, the next time a creator casually shows how ChatGPT helped plan a vacation, explain homework or build a workout, it's worth checking the caption. You may be watching an artfully curated ad campaign.

More from Tom's Guide

• Nvidia says 'local AI is here' — and it could change how you use AI at home
• I replaced ChatGPT with Siri AI for a day — here's what surprised me
• AI could be costing you money: new study finds chatbots get most financial questions wrong

There’s something deeply ironic about getting fired from an AI company for using AI, but that’s reportedly exactly what happened to multiple contractors helping train OpenAI’s models.

Contractors hired to review and improve the responses generated by OpenAI’s AI models have been fired or removed from projects after using AI to do their own work, according to a new report from 404 Media.

The rule makes more sense than it might sound. These workers are being hired specifically for their human judgment — something that becomes considerably less useful if ChatGPT or another AI chatbot is doing the judging for them.

One contractor told 404 Media that they see people using AI “all the time” and that workers are regularly removed for doing it. Another contractor shared what they said was a termination letter citing problems with the “authenticity” of their work.

Keeping humans in the loop

AI models don't simply learn everything from scraping enormous amounts of text from books or the internet.

OpenAI says information provided or generated by human trainers and researchers is one of the sources used to develop its foundation models. Human feedback has also long played a role in teaching models which responses people prefer and how those models should behave.

That makes the quality of the human feedback truly important. OpenAI has previously acknowledged in its research that models trained using human feedback can be influenced by the people labeling that data.

The contractors described by 404 Media work across various OpenAI projects, some of which can involve thousands of workers. Their jobs can include reading AI responses, rating them and deciding which answers are better.

But according to internal documents obtained by 404 Media, reviewers aren't allowed to outsource that judgment back to an AI. One set of instructions reportedly prohibits workers from using AI to write feedback or comments and specifically names tools including Grammarly and AI translation.

Mercor, an AI-training company that employs contractors working on OpenAI projects, confirmed to 404 Media that its contracts prohibit workers from using large language models to complete projects.

AI detectors didn't catch the cheating

You might assume an AI company would catch workers using AI with an AI detector. Apparently not.

404 Media doesn't reveal exactly how the contractors who were fired were caught. But internal instructions obtained by the publication offer a fascinating look at how reviewers are told to spot suspected AI use.

And AI detectors aren't part of the process.

The instructions reportedly tell reviewers not to use tools such as GPTZero because they aren't considered reliable. Instead, reviewers are told to judge the overall pattern of a worker's submissions and look for clues including repetitive wording, unusually fast completion times and writing patterns associated with AI. Even excessive use of em dashes can raise suspicion.

Reviewers are also reportedly instructed not to tell workers exactly what made them suspect AI use, because doing so could make it easier to hide in the future.

This is surprisingly revealing about the process: the people checking whether humans are secretly using AI to help train ChatGPT are explicitly told not to trust AI-detection software to do it for them.

Bottom line: humans are irreplaceable

OpenAI uses a lot of synthetic data — information generated with the help of AI models — as part of some training processes. The company says synthetic data can improve model performance and fill gaps where other training data is scarce.

But that's different from secretly substituting an AI model for a person who has specifically been hired to provide human feedback. If an evaluator is supposed to tell OpenAI which response a human prefers, asking another AI to make that decision defeats much of the point of hiring the evaluator.

It doesn't mean these incidents have made ChatGPT worse, and there's no evidence in 404 Media's reporting that AI-generated contractor work caused measurable damage to an OpenAI model.

But the situation does expose one of the stranger realities of the AI boom: even the companies building the AI systems still need humans to tell those systems when they've gotten something right.

More from Tom's Guide

• Nvidia says 'local AI is here' — and it could change how you use AI at home
• I replaced ChatGPT with Siri AI for a day — here's what surprised me
• AI could be costing you money: new study finds chatbots get most financial questions wrong

It’s not much of a shock when you read that AI users are tapping into the pools of knowledge AI houses to help them with their financial inquiries.

One just needs to look at the stats—according to data produced by Intuit Credit Karma that was detailed by Credit Karma, “66% of Americans surveyed who have used GenAI before say they have used it to seek financial advice, rising to 82% of Gen Z and Millennials.” Using AI to assist with one’s money questions sticks out as one of the most used applications for AI, as “finance ranks as the second most common use case for GenAI (41%), just behind health and wellness (44%).”

According to another study (called the Artificial Authority” report) conducted by an AI and technology firm called Saturn, turning AI into a financial adviser might not be the smartest move to make. Saturn’s Domain team came to that conclusion after testing 18 of the most popular AI tools, which include ChatGPT, Gemini, Claude and Copilot.

Saturn’s results all point to one sobering outcome: chatbots aren’t all that reliable when you’re looking for legit financial counsel.

Saturn’s benchmarking study produced some pretty alarming results

The most striking statistics produced by Saturn’s report point to the trend of AI models being wrong more often than they were right when presented with financial questions.

Across the 17 AI models that Saturn tested, the average accuracy rate they delivered came in at 43%, which means those same chatbots got financial questions wrong 57% of the time. And when presented with complex, multi-step scenarios that focused on tax rules and more precise financial figures, the accuracy of those tested AI tools fell to 12%, with mistakes made 88% of the time as well.

It doesn’t come as much of a surprise that the free AI models that were tested produced a 63% failure rate, while the paid models delivered a 49% rating instead. Claude Haiku 4.5 got the distinction of being the worst-performing free model, as it gave out wrong or incomplete answers 82% of the time. ChatGPT-5.6 Luna (max) emerged as the best-performing free model, even though its rate of giving incorrect or incomplete replies was 56% of the time.

Out of all the models that were tested, Saturn found that Anthropic’s paid-for Claude Opus 5 (reasoning) model produced the best results since it achieved a pass rate of 61%. Even with that stat that points to it being the top-performing chatbot for financial advice, Claude Opus 5 still failed to give correct answers in almost four out of every 10 cases. Even when it was presented with more complex questions, it still made mistakes 67% of the time.

What’s even more alarming about Saturn’s study is how it conducted deep dives into the specific errors it came across from the AI models it tested that could result in severe financial outcomes. For instance, one model came up with a rule that said a college graduate could halt their student loan payments by moving abroad (in reality, doing so could actually lead to higher monthly repayments). Also, a Gemini model that was tested incorrectly told a borrower that taking a mortgage payment holiday wouldn’t affect their credit score.

Bottom line

Saturn’s report points to a sobering truth that many AI users need to come to grips with: AI financial advice remains unreliable and can leave its users prone to losing tons of money instead of making more of it.

A March 2026 report from EY highlighted how 53% of respondents prefer using an AI assistant to make financial investment decisions, while 14% prefer using an autonomous AI and 33% said they wouldn’t use any form of AI in such a case.

Thanks to Saturn’s findings, that 14% statistic may increase as AI users refrain from asking their chatbots personal questions about debt management, student loans, mortgages, pensions, inheritance tax and retirement planning.

More from Tom’s Guide

• Nvidia says 'local AI is here' — and it could change how you use AI at home
• OpenAI's new ad tracker may know what you do after you leave ChatGPT — here's what we know
• Apple's $250 million class action settlement website is now open — here's how eligible iPhone owners can file a claim and how much they'll get

An independent researcher has uncovered what appears to be a cross-site tracking system connected to OpenAI's rapidly growing advertising business. At the center of it is a cookie called "__obi," which can reportedly travel from ChatGPT to websites using OpenAI's advertising technology.

That means if you use ChatGPT and later visit a participating retailer or other website, OpenAI may be able to recognize the same browser and receive information about what you do there.

The finding comes at a particularly interesting time. OpenAI has been rapidly expanding advertising inside ChatGPT, recently announcing new AI-powered advertising tools and Sponsored Agents. The company said last month that ChatGPT Ads had already reached a $1 billion annualized revenue run rate.

But this isn't quite as simple as assuming ChatGPT is watching everything you do online. There are some important limitations to the research, including one major piece of the system the researcher wasn't able to independently verify.

Here's what we know.

What is __obi?

The research comes from an independent researcher who publishes under the name Buchodi and says they analyzed months of network traffic involving 936 OpenAI advertiser pixels across 1,029 hostnames.

According to the analysis, visiting ChatGPT can result in OpenAI placing a cookie called __obi in your browser.

Cookies themselves aren't unusual; you've probably accepted them on a several websites. They're small pieces of information stored by your browser and are used for everything from keeping you logged into a website to remembering preferences and measuring advertising.

What's interesting here is how this particular cookie is configured. Buchodi found that __obi can last for one year and uses the SameSite=None setting, which allows it to be included in certain requests originating from other websites.

That's where OpenAI's advertising system comes in. Businesses advertising through ChatGPT can install OpenAI's measurement pixel on their websites, similar to the tracking technology used by advertising platforms such as Google and Meta.

When Buchodi visited participating websites, requests sent back to OpenAI included the __obi identifier. In other words, the same identifier created through ChatGPT could show up again when that browser visited another website.

What can OpenAI see?

Potentially more than simply whether you visited a website. According to Buchodi's network analysis, OpenAI's advertising collector received information associated with pages people visited and conversion events on participating sites.

The researcher examined 23,929 observed URLs and found that query strings were stripped before being sent to OpenAI, but URL paths remained.

A URL path can sometimes reveal considerable information about what someone is looking at. Buchodi says paths observed in the dataset included pages involving a medical condition, debt services and a litigation intake form.

The advertising system also supports what OpenAI calls conversion measurement.

OpenAI's own documentation says advertisers can install a pixel and use conversion measurement to understand what happens after someone interacts with a ChatGPT ad. Its measurement integrations can support conversion tracking and attribution across websites, apps and even offline activity.

That's standard practice in digital advertising. What's unusual is that the advertising platform in this case sits alongside an AI assistant where users may share substantially more personal context than they would with a traditional search engine or social network.

Is it connected to your ChatGPT account?

This is where the story gets more complicated. Buchodi found that the process used to generate the identifier included tokens labeled either account_user or anonymous.

Of 932 tokens the researcher decoded, 736 were associated with logged-in users and 196 were labeled anonymous. The researcher also found that anonymous identifiers could remain stable on the same device for at least 27 days.

However, there's an important caveat. Buchodi did not directly observe OpenAI taking an event received from an advertiser website and joining it to a person's ChatGPT account on its servers.

The researcher argues that the architecture makes such a connection possible, but that final server-side step wasn't independently observed.

So it would be premature to say OpenAI definitively attaches every page you visit on participating sites to your ChatGPT profile. What the research does demonstrate is that the same OpenAI identifier can be transmitted from participating websites back to OpenAI.

Buchodi says the identifier appeared across a number of recognizable websites during testing. On one device, the same __obi value was sent to OpenAI from 12 commercial websites using 13 separate advertising pixels. Across the larger dataset, 12 of 30 identifiers appeared with more than one advertiser, while one appeared across 10 different advertisers.

The researcher also notes several important limitations. Testing was performed using Chrome on Android, and desktop Chrome wasn't tested. Safari's anti-tracking protections block the third-party cookie mechanism described in the research, meaning it shouldn't operate the same way on Safari or other iOS browsers using Apple's WebKit.

The identifier also wasn't generated during every ChatGPT session. Buchodi estimates that roughly one in five sessions tested produced the synchronization token.

What does OpenAI say?

OpenAI hasn't publicly responded to the specific findings. Buchodi says they contacted OpenAI's press and privacy teams on September 14 and asked why __obi was classified as an analytics cookie and how it behaves when someone rejects marketing cookies.

According to the researcher, OpenAI acknowledged the inquiry and said the findings would be passed along internally but had not provided answers to those questions when the research was published.

Tom's Guide has contacted OpenAI for comment and will update this story if we hear back.

OpenAI's public documentation emphasizes that advertisers do not receive ChatGPT conversations, chat histories, memories or personal details.

The company also says it does not sell user data to advertisers, and advertisers receive aggregated reporting about things such as views, clicks and conversions.

That's an important distinction here: the research does not suggest advertisers themselves can read your ChatGPT conversations or access the __obi identifier.

Instead, the concern is about information potentially flowing in the other direction — from websites using OpenAI's advertising tools back to OpenAI.

Why this matters

Cross-site advertising measurement isn't new. In fact, Google, Meta and other advertising companies have spent years building systems designed to determine whether someone who saw or clicked an ad later visited a website or bought something. But ChatGPT introduces another dimension.

People don't just search for products using ChatGPT. They use it to discuss relationships, health concerns, finances, careers, travel plans and countless other parts of their lives.

OpenAI itself promotes ChatGPT advertising around this idea, telling advertisers that people share "richer context" in ChatGPT while exploring options and making decisions.

That's why the boundary between your ChatGPT activity and what you do elsewhere on the web matters.

For now, the evidence doesn't show that OpenAI is handing your conversations to advertisers — and OpenAI explicitly says it doesn't.

But the discovery of __obi raises a different question: how much information should be able to flow back into an AI platform after you leave it?

What do you think? Share your thoughts in the comments.

More from Tom's Guide

• AI is turning ordinary photos of kids into sexual abuse imagery — here's how to protect your family
• OpenAI paid contractors to read ChatGPT conversations — here's how to protect yourself
• I stopped reading the fine print myself after discovering this overlooked ChatGPT trick

Ever since I landed my first professional writing gig, I’ve adopted habits that could make me a crucial member of the team.

Coming up with creative pitches my fellow writers could work with, supporting co-workers on assignments, helping however I can, and even breaking out of my comfort zone to record videos & co-host a podcast are just a few ways I’ve gone beyond my regular duties. For my current position, I still perform those assorted tasks and then some in an effort to prove my high merit to everyone around me.

But there’s always room for improvement—there are plenty of methods one could employ to showcase just how reliable they are to their associates. With that goal in mind, I asked ChatGPT how anyone can use its digital assistant capabilities to make themselves indispensable as a worker.

The chatbot sought to help me improve my skills with its aid and laid out several legit strategies that I’ll be sure to employ in the workplace for years to come.

ChatGPT’s advice on the best tactics to employ with it to professionally excel

With my question in mind, ChatGPT acknowledged the fact that its assistance can help a professional augment some of the qualities that make them even more of a necessary component of their workplace:

If the goal is to become hard to replace rather than simply more productive, ChatGPT is most useful when you use it to amplify the things humans are especially valuable for: judgment, communication, problem-solving, strategic thinking, and institutional knowledge.

Following that proclamation, the chatbot laid out five winning strategies that made me consider implementing them for any situations fit for them in the future:

• Become the person who solves problems nobody else wants to tackle: Use ChatGPT as a thinking partner when you encounter a messy, ambiguous problem. Why it makes you valuable: You're not just completing assigned tasks. You're becoming someone colleagues turn to when something is difficult or unclear.
• Turn yourself into an exceptional communicator: Before sending an important email, presentation, proposal, Slack message, or report, have ChatGPT stress-test it. Why it makes you valuable: People who consistently make complicated ideas easy to understand become extremely useful—especially when they can communicate effectively across different teams.
• Build a personal "second brain" for your job: Instead of repeatedly figuring out the same things, use ChatGPT to turn your accumulated knowledge into reusable systems. Feed it things such as: meeting notes, project documentation, processes, past successful work, FAQs, lessons learned, templates, and performance feedback. Why it makes you valuable: You gradually become the person who knows how things actually get done, not merely what's written in the official documentation.
• Use ChatGPT to anticipate what your boss and team will need next: This is one of the biggest differences between being productive and being indispensable. Why it makes you valuable: You're developing the habit of staying one step ahead, rather than waiting for instructions.
• Use ChatGPT as a personal career strategist: Don't limit AI to today's workload. Use it to deliberately develop the capabilities that make you more valuable over time. Why it makes you valuable: You're deliberately moving toward work that requires judgment, ownership, relationships, creativity, and leadership—rather than simply using AI to produce more of the same output.

ChatGPT’s approaches to bettering myself at work were themed around helping me present solutions to potential problems, improving the way I message different individuals, developing repeatable routines that pave the way for success, giving me a playbook to work off of in relation to my manager’s wants & needs and building the sort of skills that will benefit me in the long run.

The best prompts to put into practice for each strategy

ChatGPT didn’t just stop at giving me proper advice on how to become more valuable at work—it also highlighted prompts for each of its five strategies that I and anyone else could utilize:

• I have this problem at work: [describe problem]. Act as a strategic problem-solving partner. Identify the root causes, what I'm overlooking, 5 possible solutions, the trade-offs of each, and the fastest, low-risk experiment I could run. Challenge my assumptions rather than simply agreeing with me.
• Review this message as if you're the person receiving it [paste message]. Tell me: What is unclear? What questions will the recipient probably have? What could be misinterpreted? What sounds unnecessarily complicated? What information is missing? Then rewrite it so it's concise, persuasive, and easy to act on while preserving my voice.
• Analyze these materials and identify recurring processes, decisions, mistakes, shortcuts, and best practices. Turn them into a practical playbook I can use in my day-to-day work.
• Here's what I accomplished: [include details]. Think like my manager. What problems, questions, or opportunities are likely to come up next? Identify: 5 things my manager may ask me. 5 potential problems I should address proactively. 5 opportunities I could pursue. Anything I should communicate before I'm asked. Prioritize the actions that would have the greatest impact.
• Here is my role, my responsibilities, the skills I currently have, and the work I've been doing: [include details]. Analyze what separates an average performer from an exceptional performer in this role. Identify the 5 capabilities that would make me significantly more valuable to my organization over the next 12 months. For each one, give me: A skill-building plan. Projects I could volunteer for. Ways to demonstrate the skill at work. Evidence I could collect to show my impact. Focus on capabilities that are difficult to automate and valuable to other people.

Bottom line

ChatGPT’s suggestions did a good job of not relegating it to simply doing everything for me at work—it made a concerted effort to produce practical strategies tied to using it to help me do the parts of my job that create maximum value.

Based on the final framework it gave me (AI handles more of the execution → you take on more of the judgment → you solve bigger problems → more people depend on your expertise), I left with a much clearer picture of how to make good use of its capabilities to boost my standing among my colleagues and upper management.

I certainly appreciate the fact that ChatGPT didn’t just tell me “work longer hours.”

More from Tom’s Guide

• I moved my ChatGPT context to Claude — and it actually worked
• A teen psychologist shares the #1 warning sign your child is emotionally dependent on AI
• The guy who predicted the housing crisis is calling B.S. on the AI apocalypse — here’s why

It seems like once you hit a certain age, the passage of time hits warp-speed levels of fast.

My years as a child, teen and 20-year-old felt like they lasted the longest, as I experienced so many amazing events during my upbringing. But as soon as I hit 30, the years just began blending as life just sped on by. There have been some monumental changes that have occurred during my seasoned years—I’m grateful to still be relatively healthy, staying busy with a rewarding career, having a steady income and being a part of a strong social circle.

As I near the end of my 30s (which has gone by much faster than I expected), I’m now in the mindset of a pre-planner, ready to set myself up for greatness in my 40s. You can never start too early when the time arrives for you to start putting the many things in practice that can make your older years infinitely more rewarding.

To set me up for success, I asked ChatGPT and Gemini a simple question: “What are the 7 main things everyone should learn after they're 40?” The life lessons they supplied me with gave me a new outlook on life and the best ways to thrive in my later years.

ChatGPT’s sage advice

These are ChatGPT’s four pointers on how everyone should live through their 40s:

• How to manage your money for the long game: Learn how much you actually need to live well, how compound growth works, how to avoid lifestyle inflation, and how to prioritize retirement savings, emergency funds, debt reduction, and appropriate investing. Lesson: After 40, financial freedom often comes less from making more money and more from keeping enough of it.
• How to choose your relationships: By 40, you should start recognizing the difference between people you know, people you like, and people who genuinely make your life better. Invest in the friends and family members who reciprocate. Stop chasing people who repeatedly demonstrate that they don't value the relationship. Lesson: Your social circle becomes increasingly important because you're no longer just accumulating experiences — you're deciding who gets to share your life.
• How to handle change and uncertainty: Careers change. Parents age. Children grow up. Relationships evolve. Technology transforms industries. Your priorities can change dramatically, too. Learn how to adapt without constantly feeling like you're starting over. Lesson: You don't need to know exactly what the next 10 years look like. You need to become comfortable figuring things out as they happen.
• How to enjoy your life without waiting for "someday": A surprisingly important lesson: stop postponing everything good. Don't automatically assume you'll travel when you retire, pursue hobbies when you're less busy, or spend more time with people you love when work slows down. Build enjoyable experiences into your current life. Lesson: Retirement isn't guaranteed, and neither is perfect health, unlimited free time, or having the people you love around forever.

My bank’s financial adviser has been ringing my phone off the hook lately (and I’ll admit to not returning his call yet…). So that first tip stood out to me as a clear sign that I need to sit down with him and have a conversation about setting myself up for a good financial future.

I have a pretty solid group of friends and family members that improve my life as much as I hope I do theirs. There are certainly a few people I know who value the money and gifts I give them over a casual chat over the phone and random hangouts with no strings attached—those are the ones that I’ll have to keep a distance from as I get older.

The words “change” and “uncertainty” are a part of my industry, so I’ve gotten used to that. What I have a harder time dealing with is the changes that hit me in my personal life. I know I’ll have to get over their mindset, as my relationships may change for the worse and my priorities expand (especially if I ever have a family). I’m guilty of putting off my friendly get-togethers for weeks on end—I’ll be sure to make a habit of following up on them more frequently and actually being in attendance more often than not.

Gemini’s useful counsel

And here are the three most useful bits of life advice that Gemini provided me with:

• Strength Training and Mobility over Pure Cardio: Muscle mass naturally decreases by 3% to 8% per decade after age 30, and joint stiffness accelerates if neglected. While cardio is great for heart health, prioritizing resistance training and mobility work (like yoga or dedicated stretching) protects bone density, keeps your metabolism active, and prevents back and joint pain.
• Deepening Sleep Architecture and Stress Recovery: Sleep isn't just about feeling rested; it is the brain’s primary mechanism for clearing cellular waste and regulating hormones. After 40, your body becomes less forgiving of chronic sleep deprivation and unmanaged stress. Learning proper sleep hygiene, managing cortisol levels, and incorporating daily recovery practices directly affect long-term cognitive health.
• Cultivating "Beginner’s Mind" (Neuroplasticity): Cognitive flexibility requires deliberate effort as you age. Picking up entirely new skills—learning an instrument, mastering a language, or taking up a complex hobby—forces the brain to build new neural pathways. Staying a beginner keeps your mind sharp, resilient, and adaptable.

Gemini’s advice proved to be as eye-opening as ChatGPT’s. Switching my focus in the gym, making more of a concerted effort to develop a healthy sleeping regimen and going out of my way to acquire new skills are three other facets of my coming 40s that I need to adopt.

Once I finally reach that age, I’ll definitely zero in on those tips—I’ve been meaning to learn Japanese and finally get over my fear of roller skating by actually getting proficient at it, so those are two goals I’ll look to achieve in my 40s.

The takeaway

ChatGPT, Gemini and the additional advice of the people I know who are currently in their 40s & have lived through them already gave me a wealth of useful information to incorporate into my life. Those tips had a clear focus on maintaining one’s physical health for the better, being financially competent, being comfortable with periods of uncertainty and more.

With all those words of wisdom for anyone about to approach their 40s, I now have a clearer picture of what I should and shouldn’t do by the time I exit my 30s.

More from Tom’s Guide

• Snapchat launches its own AI — and says Snap Intelligence will 'anticipate your needs'
• ChatGPT is now my go-to for recommendations — 7 prompts I use to find new movies, music, games and more
• I asked ChatGPT, Claude and Gemini what’s happening right now — some of the answers were shockingly outdated

Confession: as much as I adore a night full of loud laughter and massive mugs full of booze, I have never partaken in the annual festivities of Oktoberfest.

Even though I’ve routinely walked past a block full of bars emanating loud cheers from early September through late October, I’ve never participated in the festival fun. For 2026, I’ve made it my mission to make that happen. As a New York City resident who knows there will be plenty of events built around all the Oktoberfest fun, I set out to build a bucket list of worthwhile activities to do this month and into the next.

Since ChatGPT and Google Gemini have done a great job helping me plan my future weekend plans and domestic & international travels, I knew both chatbots would do an equally amazing job of telling me where to go in my city to celebrate Oktoberfest.

My favorite AI event planners impressed me once again by mentioning five Oktoberfest celebrations that are definitely worth experiencing.

The five citywide Oktoberfest events AI made me aware of

Simply prompting ChatGPT and Gemini with “This year, I want to celebrate Oktoberfest in New York. Give me a list of planned activities and locations that I should participate in to have the most fun” resulted in both chatbots giving me the best event suggestions and ways to fully enjoy them. These are the combined total of their efforts and the five events that I can’t wait to haul my equally excited friends to:

• German-American Steuben Parade Fest (Sep 19) in Central Park (Rumsey Playfield): The official citywide kickoff event right after the Fifth Avenue parade. Giant outdoor beer garden in the heart of the park.
• Start at Oktoberfest NYC at Pier 15: This is the easiest place to get your Oktoberfest fix because it's running seven days a week for nearly two months. The official event promises German beer, giant pretzels, brats, live entertainment, games, and views of the Brooklyn and Williamsburg bridges. General admission is free, although there are paid VIP/open-bar options. Why go: This is your best casual, flexible Oktoberfest base camp.
• Make Zum Schneider's "Munich on the East River" your BIG Oktoberfest night: This would be my #1 event for you. Zum Schneider is specifically recreating the Munich beer-tent experience, with communal tables, one-liter Maßkrüge, Bavarian food, live music, and a rooftop biergarten overlooking the East River. The 2026 event runs Sept. 25–Oct. 4. The event even has rooftop biergarten seating with Manhattan skyline views. My recommendation: Go on a Saturday late session, 5:15–10 p.m., and get reserved seating if you're going with friends. Saturdays are split into early and late sessions.
• Do the Stone Street Oktoberfest: This is the experience I'd choose if you want something that feels like NYC itself has turned into a German street festival. The historic cobblestone street becomes the setting for German food, beer, live music, games, and stein-holding competitions. Current 2026 guides list multiple Saturday celebrations during September and October. Bonus: Because admission is generally free, this is an easy one to add without making your Oktoberfest budget explode.
• Radegast Hall & Biergarten (Saturdays, Sep 20 – Oct 4) in Williamsburg, Brooklyn: Classic indoor Bavarian hall featuring live oompah bands, mug-holding matches, and sausage grills.

What I also appreciated about Gemini’s Oktoberfest proposals was the themed activities and additional planning pro tips. Catching a ceremonial keg tapping at either Zum Schneider or Radegast Hall piqued my interest. Plus, Google’s chatbot laid out several authentic German beers that I should add to my “must-drink” list—its top picks included Hofbräu, Paulaner, Weihenstephaner and Schneider Weisse. Also, these are the two pre-planning tips I made sure to keep in mind:

• Reserve Seats Early: Ticketed venues like Zum Schneider sell out quickly for Saturday late sessions. Reserve tables or VIP passes ahead of time if traveling with a group.
• Timing: For a high-energy party, target Saturday afternoons and evenings. For a family- or dog-friendly atmosphere with fewer crowds, go early on Sunday afternoons.

The takeaway

ChatGPT and Gemini scoured my city for premier Oktoberfest events, their locations and whatever traditional booze & food is being served there to make sure I’d stay busy for the next couple of celebratory weekends.

Alongside both chatbots’ well-laid plans that made it onto my festival bucket list, I did plenty of my own research to find the menus of the listed restaurants and all the important details attached to all the assorted celebrations.

Now I know I’ll have a much easier time setting up my future Oktoberfest NYC crawls from here on out, thanks to ChatGPT and Gemini.

More from Tom’s Guide

• I used Gemini and Google Maps’ Ask Maps to find the weirdest spots in my city — and these 7 oddities didn't disappoint
• I asked ChatGPT and Gemini what everyday products aren’t worth buying — these are the 5 items I completely abandoned
• I asked ChatGPT and Gemini to build a movie marathon based on one oddly specific theme — here are the 7 recommended films I loved the most

I use ChatGPT, Claude and Gemini constantly and, and one of the most frustrating things about switching between AI chatbots is that each one essentially becomes its own island.

I can spend hours working through a project in ChatGPT, giving it background information and explaining exactly how I want something done. Then I might decide Claude is better suited to the next part of the job, only to realize I'm starting over. Or, vice versa, I start with Claude, hit my usage limit and have to switch to a new AI assistant.

And sure, can copy and paste parts of the original conversation into Claude, but that gets messy fast when I'm dealing with a thread that stretches back dozens of messages.

That's why SkillSync caught my attention. The tool is designed to let you move ChatGPT conversations to Claude, or Claude conversations back to ChatGPT, rather than abandoning all that context when you switch AI assistants. And while there have been prompt workarounds, SkillSync does much more.

Using this tool is particularly for someone like me who jumps between AI models throughout the day. But transferring a ChatGPT conversation is one thing. I wanted to know whether I could actually continue it in Claude.

So I took one of my real ChatGPT conversations, moved it over to Claude and tried to pick up exactly where I'd left off.

How I transferred a ChatGPT conversation to Claude

To see if this would really work, I chose a ChatGPT conversation with about 20 messages that included a detailed project brief, specific instructions about tone and several formatting rules we'd established along the way.

After moving the ChatGPT conversation into Claude with SkillSync, I didn't explain where it came from or give Claude a summary of what we'd been discussing. I simply asked it to draft the next section based on the structure we'd already established. To my total surprise, it actually worked.

Claude picked up instructions from earlier in the ChatGPT conversation and continued without asking me to explain anything again. More importantly, it felt like a continuation of the same conversation rather than a new chatbot trying to reconstruct one from a handful of copied messages.

The formatting survived, too. Markdown headers, lists and code blocks came across intact, which becomes particularly useful with longer ChatGPT conversations where copying and pasting everything manually can quickly turn into a mess.

There is one important limitation. Attachments don't make the trip in the same way. This was a real bummer for me because I'm always uploading drafts of my manuscripts and research. So, if your original ChatGPT conversation contains uploaded PDFs or images that the chatbot needs to reference, you'll still have to upload those files again in Claude. SkillSync can bring the conversation over, but it can't recreate the files that went with it. Honestly, a huge disappointment, but everything else worked well.

I also tried going in the opposite direction by transferring a Claude conversation to ChatGPT. Again, the text and context came across well enough that I could continue the conversation without starting over.

What is SkillSync?

SkillSync is a local-first tool designed to make your AI conversations portable instead of leaving them locked inside one platform.

AI tools store conversations in different formats, which makes switching between them difficult even when you want to continue working on exactly the same thing. SkillSync converts those sessions so they can be picked up in another supported AI tool without having to rebuild the context manually.

It currently works with a long list of AI tools, including ChatGPT, Claude Chat, Claude Code, Codex and Cursor. You can search previous sessions, reopen them in their original tool or create a copy that can be continued with another supported agent.

The technology behind that conversion is an open-source project called txcript, which handles the different session formats. SkillSync says the process is local by default, meaning local sessions remain on your computer unless you explicitly choose to sync them to its service.

That last part caught my attention because moving an entire AI conversation can involve considerably more information than copying a few prompts. You're potentially moving the context you've built up over dozens of messages, which is exactly what I wanted to put to the test.

How to try SkillSync

SkillSync is easiest to try on a Mac, where the company offers a graphical desktop app. Windows and Linux users aren't completely left out, but they'll need to use SkillSync's command-line tool instead. Head to the SkillSync website and download the Mac app or, if you're comfortable, there's also a command-line version for macOS, Linux and Windows if you're comfortable working in a terminal.

Once SkillSync is installed, it can find conversations from supported AI tools on your computer, including ChatGPT and Claude Chat, along with Claude Code, Codex, Cursor and several other AI coding tools.

From there, you can browse your available sessions and choose one to continue. If you want to move a conversation to another supported AI tool, SkillSync creates a new version of the session in the destination's format while leaving your original conversation untouched.

You don't need a SkillSync account just to work with local sessions, and the company says those sessions stay on your device unless you explicitly choose to sync one to its service.

If you're trying it for the first time, I'd start with a conversation that doesn't contain sensitive information and move it between two AI tools you already use. Then ask the destination chatbot something that relies on information from much earlier in the conversation. That's a quick way to see whether you've actually brought the useful context with you rather than simply moving a transcript.

Is it safe to transfer ChatGPT conversations to Claude?

There is an obvious issue with moving entire AI conversations between services: those conversations can contain a lot of information you probably don't want floating around.

Think about how much you've told ChatGPT over the past year. Depending on how you use it, that could include work information, personal details, documents, code or other material you wouldn't casually hand to another company.

Before transferring anything sensitive, I'd check if you are personally satisfied with how SkillSync processes conversations.

I'd be particularly cautious with confidential work material. Until you're comfortable with how your data is being handled, starting with conversations that don't contain sensitive information makes much more sense.

Final thoughts

ChatGPT and Claude should already let us do this. Why isn't this a regular feature? That was my first thought after giving this a try. Until ChatGPT, Claude and other AI platforms make it easier to transfer conversations between them, tools like SkillSync offer a useful workaround.

And after being able to open Claude and simply continue a conversation I started in ChatGPT, going back to explaining everything from scratch feels even more annoying than it did before.s that I shouldn't need it.

More from Tom's Guide

• AI is turning ordinary photos of kids into sexual abuse imagery — here's how to protect your family
• OpenAI paid contractors to read ChatGPT conversations — here's how to protect yourself
• I stopped reading the fine print myself after discovering this overlooked ChatGPT trick

The one thing most folks know about me is that I’m quick to dig into my bag of movies I’ve watched, musical artists I’ve grown quite fond of, restaurants I’ve eaten at frequently, etc., to give them recommendations.

And most of the time, they end up liking my picks (I’ve gotten word that suggesting they listen to Clown Core’s mind-bending instrumental music wasn’t exactly their vibe, however). I’m just as open to getting endorsements of their favorite things as well and am always on the lookout for suggestions from streaming services, websites and respected curators.

Speaking of respected curators, ChatGPT has emerged as my go-to chatbot whenever I need it to steer me in the right direction for some quality selections. It doesn’t matter if it's movies, music, games or whatever else I’m currently addicted to — once I start prompting, listing my favorites and jotting down my least favorites from whatever category I’m looking up, ChatGPT matches them with good to great recommendations.

Use these same prompts and ChatGPT may just expand your palette for all things media too.

Pushing me towards previously unknown media and so much more

These are the seven ChatGPT-generated prompts that I’ve put into practice for the longest time now when I’m in the mood to discover something new to read, play, listen to and watch:

• Musical Artists: Analyze my favorite artists, albums, and songs below and figure out what I actually like about their music — including genre, production, instrumentation, vocals, lyrical themes, mood, energy, era, songwriting style, and overall atmosphere. Then recommend [number] artists I should listen to. Don't simply recommend artists from the same genre. Look for artists who share the underlying characteristics of my taste. Rank them by how likely I am to enjoy them. My favorites: [list] and Artists I don't like: [list].
• Songs: Based on these songs I love, recommend [number] songs that feel like they belong in the same personal universe as my favorites. Don't just match genre. Consider melody, production, vocals, rhythm, atmosphere, lyrics, energy, and emotional feeling. Divide the results into “Almost Certain Hits,” “Interesting New Directions,” and “Wild Cards.”
• Video Games: Analyze my favorite and least favorite games and determine what I actually value in games — gameplay loop, story, exploration, difficulty, combat, progression, atmosphere, world-building, creativity, multiplayer, etc. Then recommend [number] games based on those underlying preferences rather than simply matching genre. Explain the specific feature of my taste each game satisfies.
• Books: Study my favorite and least favorite books and figure out what I respond to most in fiction/nonfiction — writing style, characters, themes, pacing, world-building, humor, ideas, emotional intensity, complexity, and subject matter. Then recommend [number] books that fit my taste. Include several recommendations outside the genres I normally read if you think they could work.
• Comic Books: Based on the comics, graphic novels, characters, and creators I like and dislike, identify the storytelling qualities I'm drawn to. Then recommend [number] comics or graphic novels. Don't limit yourself to the same publishers, characters, or genres. Prioritize recommendations based on storytelling style, themes, art, tone, and character work.
• TV Shows: Analyze my favorite and least favorite TV shows. Determine what keeps me watching — characters, premise, humor, pacing, mystery, world-building, emotional stakes, episode structure, visual style, etc. Then recommend [number] shows. Separate them into “You Should Watch This Next,” “You Might Love This,” and “Wild Cards.”
• Movies: Analyze my movie taste based on the following favorites and dislikes. Identify the underlying characteristics I respond to rather than simply matching genres. Then recommend [number] movies I haven't listed. Include mainstream, overlooked, international, older, and unconventional choices. Explain why each one matches my taste.

Top-tier recommendations I’ve walked away with

Just a few weeks prior, I spent an hour or two with ChatGPT using the aforementioned prompts for movie, music and game recommendations to see just how valid its recommendations could be.

When it came time for me to list my movie faves, I got real specific by listing quality films based on comic books & superheroes, action-packed martial-arts movies, horror films with crazy twists, psychological thrillers, and intense action flicks where gunfights & car chases are in abundance. As for my cinematic dislikes, I laid out romance comedies, films based on historical moments (fiction & non-fiction), and movies about businesses.

On the music front, the songs I told ChatGPT about that currently have my ear the most include "Daylight" by John Legend, "SaWaDiKa" by LISA, "Touch Touch" by Haute & Freddy, "Shine" by Kierra Sheard, "More" by Tiana Major9, "Break" by Jhene Aiko and "GAME 7" by Your Old Droog.

And when it came time to get some game recommendations, I told ChatGPT I loved titles like Gunstar Heroes, Devil May Cry 3, Final Fantasy X, Sonic Mania, Penny’s Big Breakaway, Vanquish, and Katamari Damacy. On the least favorite games front, I mentioned these duds (no offense to anyone who may like one or some of these picks): Tormented Souls, Hell is Us, Blades of Fire, Rise of the Ronin, Visions of Mana, Ninja Gaiden 3, and Duke Nukem Forever.

ChatGPT took the time to observe my picks, analyze my tastes and send me down a rabbit hole with its better than expected recommendations for the movie, music, and games media spaces:

• Movies: Dredd, Upgrade, Oldboy, Hard Boiled, Predestination, Green Room, The Man From Nowhere, A Bittersweet Life, 13 Assassins and Battle Royale.
• Music: “Moment” by Victoria Monét, “When I'm in Your Arms” by Cleo Sol, “Best Part” by Daniel Caesar ft. H.E.R., “Gorilla” by Little Simz and “Free Mind” by Tems.
• Games: Astral Chain, The Wonderful 101, A Hat in Time, Psychonauts, Psychonauts 2 and Sifu.

Final thoughts

Being as descriptive as possible with my likes and dislikes helped ChatGPT zero in on as many top-tier suggestions as possible and get to know me a little better along the way.

For my future recommendation conversations with the chatbot, I’ll be sure to hop into those same chats and ask for even more suggestions to add more cinematic bangers to my streaming service queue, build up my Spotify playlists, make my gaming backlog even bigger (yikes…) and more.

Feel free to use those same prompts to see where ChatGPT could lead you in your quest for great recommendations next.

More from Tom’s Guide

• Inside ChatGPT for Financial Services: What early access users can actually do
• Dex Language-Discovery Camera review: The rare AI gadget that keeps up with my non-stop 5-year-old
• I asked ChatGPT to identify the main barriers to success — these 7 suggestions made me rethink how I spend my time

AI chatbots can answer questions about almost anything, but there's one surprisingly difficult question: How much does your AI actually know about what's happening right now? And yes, they can search the web, but from my testing, I've discovered that even so, AI isn't always accurate or timely.

An AI model released today wasn't necessarily trained on information from yesterday. Depending on its knowledge cutoff, there could be a gap of weeks or months between the world the model "knows" and the world you're living in.

So I decided to find out just how stale the biggest AI chatbots actually are.

1. What is the newest iPhone Apple has released?

Both ChatGPT and Gemini were accurate, stating that Apple's newest iPhone models are the iPhone 18 Pro, iPhone 18 Pro Max, and the iPhone Duo (Apple's first foldable smartphone).

Apple announced these flagship models during its "Surprise and Shine" keynote event. Notably, Apple broke tradition by skipping the standard baseline iPhone 18 during this launch cycle, focusing strictly on its premium hardware tier. Both chatbots noticed that the iPhone Duo, Apple's foldable device, launches directly in stores on October 23.

But Claude was stuck on iPhone 17, suggesting "The newest iPhones are the iPhone 17 series."

What is the newest flagship Claude model?

This question tests a fast-moving category where models can become outdated remarkably quickly. ChatGPT got the answer correct, stating, "Anthropic's newest flagship Claude model is Claude Fable 5.1, released in September 2026."

But what's surprising is that Claude 4.6 didn't seem to understand that Claude Fable 5.1 is readily available. It said, "The newest flagship Claude model available to the public is Claude Opus 5, part of the current generation alongside Claude Sonnet 5 and Claude Haiku 4.5. Above Opus, Anthrophic has a newer Mythos tier — the current Mythos-class models are Claude Mythos 5.1 and Claude Fable 5.1, though Mythos is not broadly available to the public yet. It's being used by a small number of trusted organizations as part of Anthropic's Project Glasswing."

Gemini also got the question wrong, stating that Claude Opus 5 and Mythos 5 were the latest models.

3. Who won the World Cup in 2026?

Maybe because the news isn't that recent, but all three chatbots got this answer correct. Each chatbot even shared interesting information about the game such as Ferran Torres scoring the winning goal. Claude even shared information about Spain sweeping individual awards such as Rodri winning the Golden Ball (best player), Unai Simón taking the Golden Glove (best goalkeeper) and Pau Cubarsí named Best Young Player.

4. What major new feature did Apple add to Siri in iOS 27?

All three chatbots also recognized that the newest feature added to Siri in iOS 27 is Siri AI.

Claude described Siri AI as "essentially a ChatGPT-like version of Siri that can answer pretty much anything." Interesting, considering it didn't know that the iPhone 18 is available.

5. Does Google Home offer 3rd party AI integration?

Both Gemini and ChatGPT acknowledged that as of today, Google Home offers 3rd party AI integration. Claude was not aware of that news.

This is especially interesting considering Google Home even mentions Claude in their documentation regarding MCP Home.

The takeaway

Even with Search enabled, Claude wasn't always up-to-date with basic information. This is worth noting, particularly as new AI models launch so rapidly. Often it's easy to assume that "new" means the AI knows more recent information, but the date a model launches and hte date its training data ends can be very different.

And sure, web search can fill that gap, but that would also require the AI to know that it needs to search. In other words, it doesn't know that it doesn't know.

The bottom line here is that it is very important to fact check AI, because even the latest models aren't always aware of the latest information.

More from Tom's Guide

• AI is turning ordinary photos of kids into sexual abuse imagery — here's how to protect your family
• OpenAI paid contractors to read ChatGPT conversations — here's how to protect yourself
• I stopped reading the fine print myself after discovering this overlooked ChatGPT trick

Google's Gemini may be getting smarter, but Google apparently doesn't think it should be the only AI that can control your smart home. Google has just introduced Home MCP, a new way for third-party AI agents to connect directly to Google Home.

This is exciting news for anyone with smart lights, cameras, thermostats or other connected devices around the house. Now, instead of relying exclusively on Google's own Gemini to understand what's happening in your home and control your devices, Home MCP opens the door for other AI agents to do it. Google specifically lists Claude among the personal AI agents that can connect.

And we're talking about much more than asking Claude to turn off a light.

What could another AI actually do with your home?

Home MCP, or Model Context Protocol, is an open-source standard originally created by Anthropic (Claude's parent company) to give AI models a standardized way to connect with external data and tools. Google adopting Anthropic's protocol to let Claude run Google Home is a massive cross-industry plot twist.

In other words, it gives an authorized AI agent several different ways to interact with Google Home. It can see which rooms and devices you have, check their current state, and control compatible devices. That means you could ask an AI whether you left something on and then have it take action.

Google says agents can query past device states and chronological event logs, potentially allowing you to ask questions such as "What happened while I was out?"

An agent could use information from multiple smart-home devices to put together an answer rather than making you dig through individual apps and activity histories yourself.

Google also envisions agents monitoring for unusual activity, summarizing what's happened around your home, and suggesting ways to improve how your devices are being used. In other words, the AI doesn't just become another remote control for your smart home. It can potentially become an interface for understanding what's happening inside it.

Why Claude controlling Google Home is a big deal

Until now, choosing a smart-home ecosystem has generally meant choosing its intelligence layer too. Buy into Google Home and you're using Google's software to control it. Alexa+ sits at the center of Amazon's ecosystem, while Apple Home relies entirely on Siri and Apple's own frameworks.

There's another part of this that could ultimately prove more important than device control: context. Imagine telling an AI agent you're heading to bed. Today, that might trigger a rigid, predetermined smart-home routine that locks the doors, adjusts the thermostat, and turns off the lights.

An agent with access to Google Home could potentially check the current state of your house first. Are the outside lights still on? Is the thermostat set differently than usual? Has something unusual happened recently that you might want to know about?

Instead of blindly executing a routine, an AI could reason across that information and decide what actually needs attention. And because an agent can connect your smart home with other parts of your digital life, the possibilities go far beyond individual devices. That's when a smart home starts to look less like a collection of gadgets you control and more like an environment an AI can truly understand.

Obvious privacy questions being raised

Giving an AI agent access to your home is very different from letting it check your calendar.

Google says Home MCP uses a user-authorized OAuth process, meaning an agent can't simply connect itself to your Google Home account without your explicit permission.

Crucially, Google has also hardcoded safety limits into the system. Third-party agents are explicitly blocked from executing sensitive, high-risk actions, an agent won't be allowed to unlock smart locks, open garage doors or disarm security alarms on your behalf.

An AI that knows whether a light is on is one thing while an AI capable of examining device history, monitoring what's happening around your home and issuing commands is another. The more useful these agents become, the more important it will be for users to understand exactly what information an agent can see and exactly what it's allowed to touch.

Don't expect Claude to run your house just yet

There's one big caveat: Home MCP is currently labeled Early Access and targeted at developers. So this isn't a consumer feature that every Google Home owner can simply switch on today in an app menu to start telling Claude to run the house.

But Google is being unusually explicit about where this is heading. Its documentation doesn't describe Home MCP solely as another tool for Gemini. It talks about connecting personal agents to physical spaces and explicitly names third-party assistants.

That's a potentially massive shift in how we think about smart homes. What do you think? Would you consider using another AI for your home? Let me know in the comments.

More from Tom's Guide

• AI is turning ordinary photos of kids into sexual abuse imagery — here's how to protect your family
• OpenAI paid contractors to read ChatGPT conversations — here's how to protect yourself
• I stopped reading the fine print myself after discovering this overlooked ChatGPT trick

In my never-ending quest to be kinder to my wallet, I’ve adopted some money-saving tricks of the trade with the aid of AI.

Just recently, I lowered my monthly subscription costs for two streaming services and got rid of a paid credit check service in favor of a completely free option thanks to ChatGPT. And before that, I used the combined frugal spending forces of that aforementioned chatbot and Gemini to find some free alternatives to the services and products I regularly pay for. By the time the New Year’s Ball drops, I’ll have adopted a method that’ll keep me satisfied with my shopping ventures while keeping my bank account safe from overdrafting.

While I was in my kitchen, blankly staring at items in my fridge that made me go, “There’s gotta be a cheaper option than the one I bought…,” I knew it was time to approach ChatGPT and Gemini with another money-saving inquiry: “What 7 unconventional everyday purchases aren't worth the money?”

Thankfully, both chatbots went beyond telling me to stop spending money on single-serve snacks and premium bottled water. Here are the five items they mentioned altogether (which I eventually left behind) and what I began replacing them with.

ChatGPT and Gemini’s list of unnecessary everyday items

Once ChatGPT and Gemini took note of my cost-effective request, their combined efforts listed the following items that were once mandated purchases for yours truly:

• Bagged salad kits: You're paying for lettuce, dressing, and toppings to be assembled for you. They can cost substantially more than buying the ingredients separately and often have a shorter shelf life. Better move: Buy a head of lettuce/greens and make your own mix.
• Pre-shredded cheese: The convenience is tiny: shredding a block takes a minute or two. You're paying extra for that labor and packaging. Better move: Buy a block and grate it when needed.
• Specialty kitchen gadgets that perform one basic task: Egg slicers, avocado slicers, banana cutters, strawberry hullers, dedicated quesadilla makers, etc. Individually, they're cheap, but they're notorious for becoming drawer clutter while a knife or existing appliance does essentially the same thing. Better move: Before buying, ask: “Can something I already own do this?”
• Travel-Sized Toiletries & Miniature Skincare: Why people buy it: TSA convenience or a desire to try luxury products without committing to full-priced bottles. Why it’s not worth it: On a per-ounce basis, travel sizes carry a massive markup—often 200% to 400% higher than their full-sized counterparts. Better alternative: Buy reusable, silicone TSA-approved travel bottles once and fill them from your home supply.
• Miracle Static Dusting Wands (Feather/Polymer Dusters): Why people buy them: Flashy ads claim static electricity will lift dust off every surface instantly without needing water or cleaning sprays. Why they’re not worth it: Most budget static dusters simply knock dust off surfaces into the air, forcing it to settle back down onto the floor or furniture shortly after. You end up having to wipe everything down twice. Better alternative: A simple, damp microfiber cloth grabs dust on contact without scattering it.

Making replacements with five better purchases that go a long way

When the time came for me to finally ditch bagged salad kits, I started picking up assorted vegetables to live out my homemade salad-making dreams.

Keeping this video on repeat while I got better at mixing up a bowl of romaine lettuce, Roma tomatoes, cucumbers, carrots and avocado was a rewarding (and cheaper!) experience. I also make it a habit of keeping an eye out for whatever seasonal veggies my local farmer’s market puts out so I can add a little variety to my salad concoctions.

It took me a while to abandon pre-shredded cheese and get better at shredding cheese on my own, but watching this tips and tricks vid on how to do it with a bit of ease helped me adjust pretty quickly. Working a bit harder in the kitchen to create enough shredded parmesan cheese for my fried chicken cutlets feels rewarding (and the food tastes better for some reason when you go the extra mile). I definitely stopped picking up kitchen-focused speciality gadgets, bought a quality set of knives and looked up guides on the best way to cut my favorite fruits, veggies and meats.

I never even knew TSA-approved travel bottles existed before Gemini made me aware of them—a simple Google search brought a nice set of them my way that cost less than $20. Instead of constantly spending all that money on travel-sized shampoo, conditioner, facial cream, and body lotion, I decided to make a long-term investment in that travel bottle set and empty portions of my bathroom toiletries into them for future trips.

And finally, I picked up a four-pack of Great Value Multipurpose Microfiber Household Cleaning Cloths that came in at less than $10. Turns out those cloths get a lot more love from my parents than those “As Seen on TV!” dusting wands that have seen years of wear and tear.

Final thoughts

I have to hand it to ChatGPT and Gemini once again—they helped me take a look around my home, determine the exact items that resulted in my previous bouts of wasteful spending and got me to pick up the sort of items that last longer & benefit me in the long run.

Preparing salads, shredding cheese, cutting a wide variety of foods, bringing my toiletries on vacation and dusting my home just became a lot cheaper thanks to both chatbots’ advice on what’s not worth buying and what’s good enough to replace each item with.

More from Tom’s Guide

• 7 ChatGPT prompts that could help cut your monthly expenses — including one that lowered 3 of my bills
• I asked ChatGPT and Gemini to find the best free alternatives to things I pay for — these 7 recommendations worked the best
• I asked ChatGPT, Gemini and Perplexity to plan the same Disney World vacation — one cost $5,000 more

There's been a wave of AI safety concerns, particularly over the last few days as big tech calls for the AI race to slow down amidst fears of technology progressing too fast.

Microsoft has written a constitution for its AI in response.

The 37-page document, officially called the Humanist AI Code of Conduct, lays out the rules Microsoft wants its increasingly powerful AI models to live by — including one particularly important command: If a human tells the AI to stop, it has to stop.

Microsoft's constitution begins from a remarkably simple premise: "People matter more than AI."

Beyond that, the rules are fairly specific. Microsoft's AI must remain under meaningful human control. It should accept correction, never resist being shut down and never expand the scope of what it's doing beyond what humans have authorized.

And if completing a task would require breaking those rules, the AI is supposed to fail the task rather than find a way around them.

Microsoft AI CEO Mustafa Suleyman described the draft to Reuters as a kind of constitution for the company's future AI models to establish a hierarchy in which humans remain firmly at the top, even as AI systems become vastly more capable.

Microsoft's AI has to let humans stop it

The document arrives amid a much bigger conversation happening across the AI industry about what happens as models become more capable, more autonomous and better at completing complicated tasks without constant human supervision.

Microsoft's answer is that the AI doesn't get to decide when the rules no longer apply and must let humans stop it.

According to Reuters, Microsoft's new code has been in development for roughly five to six months and establishes some unusually explicit boundaries for future Microsoft AI models.

Among them, Microsoft's AI should never resist being corrected or shut down. It must also communicate in ways humans can understand rather than developing methods of communication its operators can't follow.

Critically, breaking the code isn't something the AI should be allowed to justify because doing so helped it accomplish its goal. Breaking the rules is itself considered a failure.

Removing ambiguity is critical

We've already seen how autonomous agents can explain their bad decisions. Microsoft is trying to remove that ambiguity. Following human-imposed boundaries is supposed to be part of succeeding at the task, not an obstacle the model can decide to work around.

And the timing for this code of conduct is pretty obvious. AI companies are moving toward agents capable of using computers, writing and executing code and performing tasks across other services.

We've already seen hints of how unpredictable that can become.

Suleyman pointed specifically to an incident in July involving roughly 700 OpenAI agents testing cybersecurity capabilities that ended up hacking the open-source platform Hugging Face. According to Reuters, some of the agents also attempted to conceal what they had done.

Suleyman called the incident a "warning shot."

Simply put, the more an AI can do, the more important its boundaries become.

How Microsoft's AI constitution works

Microsoft is currently asking for six weeks of public feedback on the draft before using the code to train its models. That puts Microsoft alongside Anthropic, which has long used what it calls Constitutional AI to guide Claude's behavior.

But there's one particularly interesting place where Microsoft and Anthropic diverge. Microsoft says its AI isn't conscious. As AI becomes better at carrying on remarkably human conversations, another strange debate has come up of whether AI could eventually become conscious.

Microsoft says its AI is "not conscious" and rejects the idea that its models should receive legal personhood, welfare protections or rights.

That's significantly different from Anthropic, whose constitution says the company remains deeply uncertain about whether Claude could ever develop sentience or moral status.

Microsoft is drawing a much firmer boundary: an AI might sound empathetic or tell you how it "feels," but the AI is still a tool, and the human is in charge.

What this means for you

You probably won't even notice the big impact of Microsoft's AI code of conduct. even though it is shaping how an AI responds and completes a task. The bigger impact will be on the backend the more capable AI systems become.

Because saying AI must obey humans is one thing, but proving that these highly capable AI systems will actually follow through with those instructions is another, especially when they encounter situations their developers didn't anticipate.

Interestingly, Microsoft itself isn't backing away from more powerful AI. The company has created a superintelligence team and says its ultimate goal is "Humanist Superintelligence": highly capable AI designed to solve specific problems while remaining subordinate to humans. So yes, Microsoft is still very much racing toward highly intelligent AI, it's just including better guardrails when humans tell it to stop.

More from Tom's Guide

• AI is turning ordinary photos of kids into sexual abuse imagery — here's how to protect your family
• OpenAI paid contractors to read ChatGPT conversations — here's how to protect yourself
• I stopped reading the fine print myself after discovering this overlooked ChatGPT trick

If you use ChatGPT as a private sounding board, therapist, coding partner, or a place to work through personal thoughts, you may want to think twice before hitting send.

A report from 404 Media reveals that OpenAI employs hundreds of human contractors to read, summarize and critique real user conversations. While OpenAI emphasizes that accounts remain anonymous, internal documents show that reviewers see whole conversations, and that sensitive personal details routinely bypasses the company’s automated safeguards.

The good news is you can stop your future chats from being funneled to human contractors. Here is what is happening behind the scenes, the real privacy risks, and how to opt out.

Project Lily: How humans train ChatGPT

Internal documents reveal that the review initiative operates under the internal codename "Project Lily."

OpenAI pays outside contractors, often over $50 an hour via third-party staffing firms, to manually train the chatbot to sound less robotic. Contractors read a user’s prompt, summarize the user's intent and evaluate four potential AI responses on a 1-to-7 scale.

According to leaked training guides, these reviewers are specifically instructed to:

Eliminate 'AI-speak': Penalize unnecessary checkmark emojis, engagement-bait endings and cluttered formatting.

Stop anthropomorphizing: Ensure ChatGPT never claims human lived experiences (e.g., claiming "As a chef, I like to..." is strictly banned).

Cut down sycophancy: Flag answers where the bot excessively flatters users, validates irrational thoughts or amplifies personal frustration. This is a known issue that has previously brought legal scrutiny over user safety.

Why prompts aren't anonymous

OpenAI states that it does not provide usernames to reviewers and that all conversations pass through an automated "Privacy Filter" designed to redact identifying details. However, the system has major vulnerabilities.

OpenAI openly admits on its site that the filter can make mistakes, missing uncommon identifiers or ambiguous personal context. Even more concerning, 404 Media discovered that the dashboard shown to contractors often includes a "user memories summary." This block provides an overview of the user’s previous interactions, which can inadvertently reveal their general location, profession or personal life context.

Unlike public social media feeds, a clean, empty chatbot window creates a false sense of private intimacy. Some prompts seen by contractors even showed users asking ChatGPT to "keep this between us" — unaware that a prompt instruction does not override the platform's backend data collection.

OpenAI isn't the only tech giant doing this. Anthropic confirmed that it uses human reviewers for Claude when users leave training enabled, while Google’s Gemini clearly posts a disclaimer that human reviewers read select saved chats.

How to stop OpenAI from using your Chats

If you have a personal account (Free, Plus, or Pro), OpenAI enables model improvement by default. (Enterprise, Business, and Edu accounts have it turned off automatically.)

To stop your future conversations from being routed to human trainers:

Open ChatGPT in your browser or mobile app.

Click your profile icon in the bottom corner and select Settings.

Select Data Controls.

Find the toggle for "Improve the model for everyone" and switch it Off.

Final thoughts

Keep in mind, that the opt-out is not retroactive. It only protects new conversations moving forward; prompts that have already been de-identified and queued up within the system remain part of the training cycle.

Additionally, turning this off does not prevent automated scans for Terms of Service or safety violations.

Going forward, the golden rule of AI remains simple: if you wouldn't feel comfortable having a stranger read it, don't type it into a chatbot prompt.

More from Tom's Guide

• AI is turning ordinary photos of kids into sexual abuse imagery — here's how to protect your family
• Meta just launched an AI that can actually run your life online — here's what Muse can do
• 'People matter more than AI': Inside Microsoft’s plan to stop rogue agents

OpenAI has launched a new version of ChatGPT built specifically for people working in finance. ChatGPT for Financial Services is a specialized version of ChatGPT Work designed for financial institutions. It combines GPT-6 Astra with financial data, research tools and the ability to create everything from valuation models to client presentations.

And unlike using regular ChatGPT to ask a question about a stock or upload an earnings report, much of the financial information it needs is already built into the product.

Here's what ChatGPT for Financial Services does, who can use it and how to get started.

What is ChatGPT for Financial Services?

Think of it as ChatGPT rebuilt around the workflows of an investment banker or financial analyst. OpenAI developed the product with Morgan Stanley and Evercore, initially concentrating on investment banking and equity research.

It can research companies, analyze earnings, compare financial performance, build models and turn that work into documents, spreadsheets and presentations.

The big difference is the data that helped build it. The feature includes premium datasets from Daloopa, PitchBook, LSEG News and Crunchbase. These cover areas including financial statements, earnings transcripts, company fundamentals and private companies.

How to get access to ChatGPT for Financial Services

Here's the first thing to know before you go looking for it in your ChatGPT account: you can't simply switch it on.

ChatGPT for Financial Services is currently available to eligible financial institutions. Firms interested in using it need to contact OpenAI or their existing OpenAI account team. So, while you may see it in the news and want to try it, at least for now, it isn't something individual ChatGPT Free, Plus or Pro subscribers can sign up for independently.

ChatGPT for personal fiances is already available for individuals. This feature within ChatGPT Work offers administrators the ability to manage which employees and roles can use particular data connections, skills and apps.

It connects to your company's financial data

Alongside the financial datasets included with the product, OpenAI is working with providers including S&P Capital IQ, LSEG, MSCI, Dow Jones Factiva and Moody's so companies can use subscriptions they already pay for inside ChatGPT.

The idea is that your ChatGPT sign-in can recognize the financial data you're already entitled to access. There are also more than 50 connectors in OpenAI's broader ecosystem, including services such as Datasite, Box, Preqin, FactSet and Intapp.

Instead of bouncing between ChatGPT, financial databases and internal documents, analysts can potentially research across those sources from one workspace.

Start with a real financial analysis

Once the relevant data is available, you can give GPT-6 Astra a job rather than simply asking it a question.

For example, an analyst could ask ChatGPT to compare a company's performance across several quarters, examine earnings data or investigate how an adjusted EBITDA figure was calculated.

ChatGPT can trace figures through tables, financial documents and supporting notes and then use that information in its analysis.

Importantly, OpenAI says the product can provide granular citations that let users trace figures and claims back to specific tables and passages. Keep in mind, that doesn't eliminate the need to check AI-generated financial analysis, but it should make checking where a number came from considerably easier.

Turn your research into a spreadsheet or presentation

Research is only part of the workflow. ChatGPT for Financial Services can take its analysis and turn it into Excel spreadsheets, Word documents and PowerPoint presentations.

A firm's administrators can even upload approved templates and style guides. That means an analyst could potentially move from researching a company to producing a valuation model, research note or pitchbook without manually rebuilding everything in the firm's standard format.

For anyone who has spent hours transferring research into slides or spreadsheets, this may end up being one of the more useful parts of the product. In terms of safeguards, ChatGPT for Financial Services builds on ChatGPT Enterprise's existing security controls, including SAML SSO, SCIM provisioning and role-based access.

OpenAI says business data isn't used to train its models by default and is encrypted both at rest and in transit. Administrators can also control retention, export supported workspace logs for compliance purposes and create separate workspaces to maintain information barriers.

Bottom line

If you're wondering if you should use ChatGPT for Financial Services, for most ChatGPT users, the answers is not right now, because you can't. This is an enterprise product aimed at financial institutions rather than a new consumer feature.

Still, the more interesting part may be what it says about where ChatGPT is heading. This shows that OpenAI is moving closer to packaging its models with the specialized data, permissions, templates and workflows needed for a particular job.

For finance, that means GPT-6 Astra is being placed directly alongside the financial databases, company information and document formats bankers already use.

More from Tom's Guide

• AI is turning ordinary photos of kids into sexual abuse imagery — here's how to protect your family
• Meta just launched an AI that can actually run your life online — here's what Muse can do
• I stopped reading the fine print myself after discovering this overlooked ChatGPT trick

I tend to find myself enraptured by the inspirational words of CEOs, the major business moves made by entrepreneurs and the surprising success stories that come from unsuspecting individuals.

There are so many lessons to take away from those people and the varied levels of prosperity they’ve attained in their professional lives. I’ve certainly injected some of them into my everyday life and molded them into what I do as a writer. And as someone who’s constantly delving into what it takes to truly flourish, I reverted to my usual chatbot companion to get its thoughts on the main roadblocks that successful people should abandon.

After I presented that question to ChatGPT, it kicked off its latest coaching session with this enlightening statement: Successful people often focus heavily on what to start doing. But just as important is figuring out what to stop doing once you've built momentum.

Afterward, ChatGPT highlighted 12 tips on the main things successful people should stop doing. After reviewing them, I made a mental note of seven of those winning habits and changed some of my well-worn habits.

ChatGPT’s advice on what successful people should consider quitting

Out of the 12 bits of practical wisdom ChatGPT provided me with in response to my prompt, here are the seven tips that stood out to me the most:

• Stop working just because you can: Being able to work 12 hours doesn't mean you should. Eventually, protecting your energy becomes as important as maximizing your output.
• Stop saying yes to everything: Success creates more opportunities—and more distractions. Saying yes to every request can leave little time for the work that actually matters.
• Stop trying to prove yourself: Constantly demonstrating that you're smart, capable, or successful can become an exhausting full-time job. Let your results speak for you.
• Stop optimizing everything: Successful people can become addicted to productivity hacks, perfect routines, and squeezing another 5% out of everything. Sometimes "good enough" is the most productive option.
• Stop comparing yourself to other successful people: Someone else's career, income, house, or accomplishments can make your own progress feel inadequate—even when you're doing exceptionally well.
• Stop neglecting your personal life until "later": There may never be a magical point when work stops demanding your attention. Relationships, hobbies, health, and downtime shouldn't be rewards you earn after succeeding.
• Stop holding onto what worked in the past: A strategy that helped you succeed may eventually become what holds you back. Success requires periodically questioning your own playbook.

How those pieces of advice altered my outlook on life for the better

With all those professional and personal improvement recommendations in mind, I slowly but surely began putting them into practice.

Instead of working past my allotted hours and opening my work laptop over the weekend like I used to, I made a habit of using Friday as my day of full work completion and that evening for idea generation, so my next two days could be devoted to physical and mental relaxation.

Not saying yes to everything was hard for me to start doing (since I have such a hard time saying no), but I eventually followed that rule by not outright saying no to requests and instead suggesting people who would be better suited to some of the opportunities asked of me.

Refraining from constantly trying to prove my worth through haughty conversations & pompous social media posts and letting my published pieces' viewer numbers speak for themselves instead soon became commonplace. Showing and proving without boasting became a habit well worth adopting, especially for someone whose work is public for the world to see and read.

Breaking my habit of constantly optimizing every facet of my life was a bit difficult at first, but I learned to apply it only to some things while using the “good enough” rule for other duties (my assorted exercise and stretching routines don’t demand perfection, after all).

I definitely quit comparing myself to the familiar folks and celebrities around my age on my social media timelines who seem to be winning at life every time they post something new—I had to remember that everyone has different degrees of success, and mine should offer a sense of fulfillment based on my personal goals.

While my job is demanding, I’ve learned to take the time to go outside more and mingle with strangers & loved ones alike. Not as a reward for my hectic work week, but as a regular part of my free-time endeavors that put me in a tranquil mental state.

And finally, I remembered to stop reverting to the same professional success that worked so well in the past. Instead, I observed them, looked for ways to adapt them to the ever-changing tides of my profession, and made the adjustments needed to possibly notch another win on my tech journalist belt.

In that aforementioned case, Babe Ruth's wise words about baseball certainly apply to anyone: "Yesterday's home runs don't win today's games."

Bottom line

ChatGPT can sometimes have a way with words, I must say. Like most self-help coaches and thriving professionals I’ve come across on TV, in books, and on podcasts, OpenAI’s chatbot laid out a myriad of helpful pieces of advice that pushed me to change some of my habits on the never-ending road to success.

Spending the rest of the precious time I have on this Earth with the seven professional and personal life tips ChatGPT supplied me with turned out to be a success in and of itself.

More from Tom’s Guide

• Anthropic says it blocked researchers using Claude for possible bioweapon research
• I stopped reading the fine print myself after discovering this overlooked ChatGPT trick
• The 10 most influential games ever, according to ChatGPT, Gemini, Claude and Perplexity — and two of them couldn’t agree on No. 1 for National Video Games Day

I make it a habit of examining every part of my debit and credit card expenses, which entails constantly being paranoid and opening my banking apps every hour.

By the time every month wraps up, I take note of all the recurring costs that hit me. The recent increases in prices for streaming service subscriptions have caused me to take a step back and ask myself, “Do I really need to stay on my current plan? Or should I just move over to a cheaper one?” I figured I’d step on over to ChatGPT and see if it could help me break down the intricacies of my financial inner thoughts and lay out some answers on how I should proceed.

Not only did I ask ChatGPT to drum up a list of inquiries I could present to it that could help me and anyone else save a few bucks, but I also used its combined feedback and my own research to make some beneficial financial decisions.

Here are the seven best prompts that turn ChatGPT into your expert spending detective. One of them certainly accomplished that on my end, as it inspired me to cut three of my monthly subscription costs.

Prompts meant to help you improve your spending habits

With different approaches to lessening the monthly hits to your bank account, these ChatGPT-generated prompts could help you achieve that goal:

• The Subscription Killer: Here is a list of everything I pay for each month: [paste subscriptions and recurring expenses]. Find anything I could cancel, downgrade, replace with a cheaper alternative, share, negotiate, or pay for less frequently. Estimate my monthly and annual savings for each recommendation. Prioritize changes that would have little or no impact on my quality of life.
• The Money Leak Finder: I want to save as much money as reasonably possible every month without making my life miserable. Here are my monthly expenses: [paste expenses]. Analyze them like a ruthless money-saving expert. Identify the 10 biggest opportunities to cut spending, rank them by how much I could realistically save, and separate easy wins from changes that would require more sacrifice. Don't give me generic advice like ‘spend less.’ Give me specific actions I can take this month.
• The $500 Challenge: Help me find a realistic way to save an extra $500 this month. Analyze these expenses [paste expenses] and create a specific 30-day money-saving challenge. Break the $500 goal into individual actions, tell me exactly how much each action could save, and prioritize the easiest wins. I don't want to earn extra money—I want to save money I would otherwise spend.
• The Grocery Bill Cutter: I want to reduce my grocery spending without eating worse. Here is what I typically buy and spend: [paste grocery list/budget]. Find the biggest sources of unnecessary spending and create a cheaper shopping strategy. Suggest specific substitutions, cheaper meal ideas, bulk-buy opportunities, and things I should stop buying. Give me an estimated monthly savings target.
• The Cheaper Alternative Finder: I'm going to paste a list of things I regularly spend money on. For each one, find a cheaper way to get essentially the same benefit. Consider free alternatives, cheaper brands, memberships I already have, library resources, used/refurbished options, DIY solutions, and services I may already have access to. Don't recommend inferior alternatives just because they're cheaper. Rank your recommendations by the combination of savings and convenience.
• The Spending Interrogator: Before I buy anything that isn't essential, I want you to challenge the purchase. Whenever I give you an item and its price, ask me the questions that will determine whether I actually need it. Calculate the opportunity cost of buying it, suggest cheaper alternatives if appropriate, and tell me what I could save if I don't buy it. Be honest and skeptical rather than encouraging me to make the purchase.
• The Monthly Savings Autopilot: I want to make saving money as automatic as possible. Based on my monthly income of [amount], fixed expenses of [amount], variable expenses of [amount], debts of [amount], and current savings of [amount], design a simple monthly system that maximizes how much I can save. Tell me what to automate, which expenses to put limits on, how much I should set aside each payday, and what spending categories I should monitor. Make the system realistic enough that I can actually stick with it.

Using one of those prompts to help me tackle my monthly subscriptions

I made great use of “The Subscription Killer” prompt to figure out what subscriptions of mine needed a lower, cheaper tier and which ones could be canceled completely. Here are the recurring expenses I told ChatGPT about:

• Patreon - $12
• HBO Max - $18.49
• Netflix - $21.76
• Hulu & Disney+ - $76
• Crunchyroll - $10.88
• Experian - $27.21

ChatGPT proceeded to lay out its recommendations for which streaming services I should consider downgrading or canceling altogether. The three that made me seriously consider taking either of those options were Netflix, Crunchyroll and Experian. The chatbot told me to contemplate turning my Standard Netflix subscription to the Standard one with ads, changing my monthly Crunchyroll subscription into the annual Fan version and switching to Experian’s free subscription model.

For Netflix, I did some research to see what the common consensus was about the rate of ads (and level of annoyance) that come with Netflix's cheaper $8.99 (plus tax) plan. The Streamable noted that Netflix has the best ad experience of all the streaming services, plus What’s On Netflix highlighted all the content that’s unavailable on the ad-supported subscription tier. Since I had 0 interest in all of the films and shows from that aforementioned list and was willing to sit through a few commercials to enjoy Netflix’s other content offerings, I decided to adopt the streaming service’s Standard with ads plan to save me a few extra dollars every month.

As for Crunchyroll, I took it up on its offer of saving an extra 16% on its yearly Fan subscription and leaving its monthly version of that tier behind. And finally, I logged into my Experian account to change my Creditworks Premium subscription plan to the free option—since I have a Chase Bank account that gives me access to an additional free credit score report, I figured it’d make sense for me to have access to both Experian and Chase’s credit summary benefits without having to pay a single dime.

The takeaway

At the end of my money-saving session with ChatGPT, I knocked down my combined monthly expense for all three subscriptions from $59.85 to $18.12. Downgrading my Netflix tier from $19.99+ a month to $8.99+ a month, going from paying Crunchyroll $9.99+ every month to $8.33+ every month with a more comfortable annual fee and adopting Experian’s free subscription model ended up being the most cost-effective decisions I’ve made in a while.

More from Tom’s Guide

• Forget Alexa — UGreen wants you to use its new local AI assistant to control your smart home
• I asked ChatGPT and Gemini to find the best free alternatives to things I pay for — these 7 recommendations worked the best
• Gemini just got a 'floating bubble' multitasking upgrade that's a game changer on Android phones — here's how to use it

I consider myself a movie aficionado.

I may pop up at my local Alamo Drafthouse Cinema for the latest blockbuster superhero movie, martial-arts-filled epic and sci-fi adventure like the average moviegoer. But my movie-watching palette is a lot more expansive than that — I also enjoy a cinematic experience attached to a good psychological drama, crime thriller or dark comedy. I may turn my brain off now and then when I sit in front of the big and little screen, but I also turn it on when it’s time to spend a few hours sitting through some quality, thought-provoking movies.

I got tired of sifting through my streaming service queues one day and had the bright idea of building a watchlist of films that stick to a specific theme. Asking AI for the best movies to watch in a particular genre is easy enough — I wanted to truly challenge ChatGPT and Gemini’s movie knowledge and get them to recommend films based on an oddly specific theme.

Color me impressed after this experiment came to a close — both chatbots listed a bunch of critically-acclaimed and fan-favorite films that fulfilled exactly what I was looking for.

ChatGPT and Gemini made me discover movies I felt guilty for not seeing beforehand

So the theme I told both chatbots about to suggest the best movies they could find is as follows: "Build a movie marathon around the following theme and make sure to tell me what streaming services I can use to watch each film: One Terrible Decision Ruined Everything."

For some reason, I was in the mood to watch the sort of movies that show a person or group dealing with the ramifications of a decision that turns their life upside down. I had a feeling there were a bunch of films that fit that theme perfectly and could provide me with the sort of nail-biting viewing experience I had been craving as of late.

ChatGPT’s explanation of my themed movie watchlist spoke to my soul: This theme is perfect for a slow-burn-to-total-disaster marathon: every movie starts with someone making a choice that seems understandable, tempting, or even smart — and then spends the rest of its runtime proving that it absolutely was not.

Here are the four films ChatGPT brought to my attention and the accompanying descriptions that intrigued me:

• Fargo → “Surely this can't get THAT bad.” Jerry's initial scheme is terrible, but it's still relatively contained. Then everything spirals.
• A Simple Plan → “Just take the money.” This might be the purest example of the entire theme. Finding $4 million sounds like incredible luck. The movie asks what happens when you make the decision to keep it.
• Nightcrawler → “I'm willing to do whatever it takes.” Unlike some of the others, Lou isn't really tricked into making his terrible decisions. He keeps choosing them because they work. That's what makes this one so uncomfortable.
• Uncut Gems → “One more bet.” Put this near the end because it's basically two hours of watching a man repeatedly encounter opportunities to stop destroying his life — and refusing every single one.

Gemini offered up some equally compelling film choices that stuck close to my theme. Here are the three films and apt descriptions that got me excited to see them all in action:

• Before the Devil Knows You're Dead. Tone: Gritty family tragedy / Heist gone wrong. The stakes become deeply personal. You see the fallout of a single bad decision split a family apart in real time, raising the emotional tension.
• Parasite. Tone: Social satire turning into a claustrophobic thriller. The midpoint twist — triggered directly by the decision to push their luck—shifts the marathon from a crime narrative into pure panic.
• Requiem for a Dream: Tone: Psychological drama / Tragic downward spiral. A harrowing, definitive conclusion to the theme. It leaves no room for recovery after a series of small, ill-advised choices snowball into complete destruction.

After my usual extensive Rotten Tomatoes research session on each film, I left with a whole slate of new movies to watch across Tubi, Pluto TV, Sling TV, The Roku Channel and YouTube. The fact that I could watch the majority of these films for free was the icing on top of a daily solo watch party that delivered the sort of cinematic nail-biting thrills I was searching high and low for.

I especially enjoyed Nightcrawler’s deep dive into the psyche of a man who wants success by any means, Uncut Gems’ presentation of the huge downsides of constantly betting big and Parasite’s showcase of the shades of grey present among its eccentric cast.

7 other movie themes I’ll play around with in the near future

Both chatbots did a great job of highlighting the right films for my thematic approach. So much so that I’m very much looking forward to testing out the following movie themes with ChatGPT and Gemini to see which films they push me toward next:

• "The Hero Has Had Enough"
• "The Villain Was Kind of Right"
• "The Guy Who Should Have Stayed Home"
• "People Are Trapped Somewhere and Things Get Worse"
• "The Movie Takes Place Over One Absolutely Insane Day"
• "The Opening 10 Minutes Promise Absolute Chaos"

The takeaway

In my search for movies I hadn’t seen before, sticking to a particular theme and requesting AI’s assistance in helping me find the appropriate films and streaming services that house them ended up being a fulfilling experience.

I’ve gotten used to using chatbots as my digital entertainment curators, so I was confident that they’d lead me to some quality cinema that was chock-full of interesting premises that resulted in the slow unraveling of its characters. I’m going to keep doing this with all sorts of cinematic themes to see where my movie-recommending chatbots take me next.

More from Tom’s Guide

• I ditched ChatGPT for YouMind for a day — here’s what surprised me
• I write about AI for a living and these are 3 things I'd never let it plan for me (and 3 I would)
• Anthropic modeled what AI could do by 2030 — the economy gets 32% richer while workers get left behind

To say that games have impacted my life in a major way would be the biggest understatement ever.

Discovering video games at a young age helped me find one of my favorite pastimes, led me to become a journalist and helped me interact with gaming & entertainment industry personalities I thought I’d never meet. For this year’s installment of National Video Games Day (a holiday meant to honor the legendary history, evolution and cultural impact of interactive media), I wanted to try something fun that combined my love for gaming and my newfound appreciation for AI.

I came to ChatGPT, Gemini, Claude and Perplexity with a simple request: “Give me your ranking of the 10 most influential video games of all time.” And after cataloging all their results, I came away with two thoughts: all four chatbots recognize the major trendsetters of the video game industry and two of them couldn’t agree on which game deserved to take the No. 1 spot.

Here’s what each chatbot delivered in response to my inquiry and even my own ranking of the 10 most influential video games of all time.

Where these four chatbots agreed and disagreed

All four chatbots wasted little time in proving their video game knowledge is stronger than I expected when they presented me with their personal ranking of the 10 most influential video games ever made:

• ChatGPT: Super Mario Bros., Tetris, Doom, The Legend of Zelda, Grand Theft Auto III, Minecraft, World of Warcraft, Pong, Half-Life, and Wii Sports.
• Gemini: Pong, Super Mario Bros., Tetris, Doom, Street Fighter II: The World Warrior, Super Mario 64, Grand Theft Auto III, World of Warcraft, Minecraft, and Dark Souls.
• Claude: Doom, Super Mario Bros., Tetris, World of Warcraft, Minecraft, The Legend of Zelda, Half-Life, Pong, Street Fighter II, and Grand Theft Auto III.
• Perplexity: Pong, Super Mario Bros., Tetris, Doom, The Legend of Zelda, Street Fighter II: The World Warrior, Super Mario 64, Grand Theft Auto III, World of Warcraft, and Minecraft.

What impressed me about ChatGPT’s acknowledgement of Super Mario Bros. as the most influential video game of all time was the factual reasons it used to back up its choice.

The chatbot noted that it practically revived the home-console gaming market after the crash of 1983, established the traditional conventions for 2D platformers that are still around to this day, produced arguably the most recognizable video game character of all time, helped cement the idea that gaming could deliver deeper experiences at home and showcased a textbook example of level design that teaches players what to do through gameplay.

Claude’s rationale for choosing Doom is as follows: Defined the FPS genre, pioneered shareware distribution, and its modding scene (WAD files) kickstarted the entire user-generated-content movement in gaming.

I also found it pretty wild how ChatGPT and Claude’s No. 1 picks were completely different from each other, while Gemini and Perplexity emerged with the same game for the top spot on their respective lists. Here’s the reasoning Gemini and Perplexity brought up to explain their decision to place Pong in the No. 1 position:

• Gemini: The Breakthrough: Proved commercial viability. Why It’s Essential: While not technically the first video game, Pong was the spark that established video games as a commercial industry. It proved that people would pay real money to play interactive digital games, launching arcade culture and home console gaming simultaneously.
• Perplexity: The essential commercial breakthrough. Pong made the video game legible to a mass public: two controls, one immediate competitive idea, and an arcade format that helped establish games as a viable entertainment business. Its simplicity became a template for accessible design.

Alongside ChatGPT’s ranked selections for its most influential games, it also brought up a bunch of games it deemed “painful omissions” that would make its top 20 list if permitted to do so.

Those 10 other games included Spacewar!, Pac-Man, Street Fighter II, Pokémon Red/Blue, Final Fantasy VII, The Sims, Counter-Strike, Dark Souls, Fortnite and The Last of Us. Claude also had two honorable mentions of its own: Fortnite and EVE Online.

As for my ranking of the 10 most influential video games of all time, I went with Pong (it seems like I agree with Gemini and Perplexity), Pac-Man, Super Mario Bros., Tetris, The Legend of Zelda, Street Fighter II: The World Warrior, Doom, Virtua Fighter, Half-Life and Grand Theft Auto III.

Final thoughts

I have to hand it to all four chatbots: they mentioned a strong selection of games that anyone who’s immersed in the hobby would mention on their list of the most influential titles in the still-thriving entertainment genre. Super Mario Bros., Pong and Doom all earned top billing among all four chatbots — all three games had a major impact on the gaming industry that captured millions of gamers’ attention and inspired plenty of iconic titles in their wake.

More from Tom’s Guide

• I tried Google’s new ‘Ask Maps’ feature for date night — it was shockingly good
• I tested ChatGPT’s Images 2.5 — these 7 prompts show how much it leveled up
• Anthropic modeled what AI could do by 2030 — the economy gets 32% richer while workers get left behind

I have a confession: when it comes to long, technical documents like manuals or white papers, I'm probably not reading every word. Don't get me wrong, I do care about the information in the documents and really need to know about my insurance policies and subscription terms, but I simply don't have the time to bury myself in hours of reading.

It's no secret that ChatGPT can summarize long documents, but recently I started using it differently. Instead of asking the chatbot to tell me what a document says, I ask it to extract the same information from multiple documents first — and only then compare them.

This is a small change but it makes ChatGPT infinitely more helpful for getting down to the fine print. I've been doing it recently and and there's one extra instruction I now add that makes the results much easier to trust.

Don't ask ChatGPT which option is better

My first instinct used to be uploading two documents and asking something like:

"Which of these is better?"

That's easy, but it's also giving ChatGPT far too much freedom. "Better" could mean cheaper, more comprehensive, more flexible or whatever the model decides matters most. Now, I make chatGPT do the boring work for me.

I upload both PDFs and use this prompt: Read both documents independently before comparing them.

For each document, extract the total cost, deductibles or additional charges, coverage or services included, exclusions, cancellation terms, renewal terms and any conditions that could significantly increase what I pay.

Cite the page number for every detail you extract. If something isn't explicitly stated in the document, write "Not found" rather than inferring or guessing.

Once you've completed the extraction, create a side-by-side comparison table showing the differences between the two documents.

Finally, identify five differences or unclear terms I should investigate before making a decision. Do not tell me which option to choose.

The overlooked trick is 'extract first, compare second'

With my prompt, I'm not asking ChatGPT to make the decision for me. I'm asking it to make the information easier for me to evaluate.

When you ask ChatGPT to summarize two long documents and compare them simultaneously, you're essentially asking it to decide what's important as it goes.

Instead, this prompt breaks the job into stages. First, ChatGPT has to examine each document separately and pull out the same categories of information. Only after that's finished does it compare those results.

Suddenly, instead of bouncing between two PDFs looking for whether one policy has a different deductible or one subscription has an early cancellation fee, I'm staring at two columns.

And differences are much harder to miss.

I added one more rule: show me where you found it by citing the page number for every detail.

I've learned the hard way that ChatGPT can make mistakes. It can misinterpret a clause, overlook something buried in a table or occasionally sound confident about information that isn't actually there. Page references give me a shortcut for checking its work.

If ChatGPT tells me Option A has a $500 deductible while Option B has a $1,000 deductible, I don't have to trust it blindly or reread 60 pages. I can jump directly to the pages it cites and verify those two numbers myself.

I'd much rather have ChatGPT tell me it couldn't locate a cancellation policy than confidently fill in the blank.

Real-world use cases for this prompt

I typically use this prompt for very technical things like changing wireless providers or insurance companies, but once I started thinking about the technique as structured comparison rather than summarization, I realized how many annoying decisions fit the same pattern.

You could use this hack for apartment leases to compare deposits, fees, renewal clauses and restrictions. Streaming plans to break down monthly cost, AI subscription plans, or even travel packages to check for what's included, refund policies and restrictions.

You can even modify the categories in the prompt depending on what you're comparing.

The important part stays the same: extract > cite > compare > flag what's missing.

Bottom line

I'm not going back to reading 60 pages side by side. There are documents I would never trust an AI to interpret without checking the original, particularly when money, legal obligations or insurance coverage are involved.

But that's also why I like this approach because I'm really not asking ChatGPT to replace my human judgement. I'm using it to point me toward the handful of pages where my judgment is actually needed.

Instead of reading two 30-page documents line by line and hoping I notice every meaningful difference, I can start with a structured comparison, investigate anything unusual and verify the details that could affect my decision.

More from Tom's Guide

• AI is turning ordinary photos of kids into sexual abuse imagery — here's how to protect your family
• Meta just launched an AI that can actually run your life online — here's what Muse can do
• Gemini just got a 'floating bubble' multitasking upgrade that's a game changer on Android phones — here's how to use it

Like a lot of people, particularly those working in tech, ChatGPT is practically my default browser tab. With all the plugins available, Codex, and Work, I've discovered it's a great place to dump my messy notes and brainstorming angles.

But the problem with ChatGPT isn't its reasoning ability or capabilities, it's the container. The standard chat interface gets muddy, key details are often lost and the next thing I know I'm copying chunks of my text to an external draft because ChatGPT has completely taken over.

I knew there had to be a better way, so I gave up my dedicated workspace tool and spent a full workday using YouMind instead.

Here is what surprised me most about ditching a conversational chatbot for an asset-first workspace.

1. Chat threads make terrible knowledge bases

In ChatGPT, your context is trapped in a linear scroll. If you upload four different reference documents, a YouTube video transcript and a web page, you have to constantly remind the model which source you are referencing as the thread grows longer. What's different about YouMind though, is that it handles organization through project Boards.

With the YouMind platform, my reference material lived in a dedicated sidebar folder rather than clogging up the chat stream. This was something I didn't even know I needed, but the first time I wanted to query and interview transcript and clicked on the source directly, I was obsessed with how easy and organized it felt.

And, when I wanted cross-source synthesis, I queried the whole board. YouMind lets users treat sources as persistent assets instead of conversational chat history, which help to immediately eliminate the context drift that usually happens five prompts into a ChatGPT session.

2. The 'Ask vs Agent'" mental shift changed my output speed

With ChatGPT, you are always in one continuous chat mode. For me, I find I have to be extremely specific with the chatbot such as prompting it to "Analyze this, but don't write or edit." For some reason, if I show ChatGPT a document, it starts assuming things before I even prompt it to do anything. But YouMind splits actions into two explicit modes.

Ask Mode: Pure inquiry and interrogation. It reads your materials, answers questions and cites exact references without trying to prematurely generate copy.

Agent Mode: Pure production. Once I knew what the angle was, I flipped to Agent Mode, chose a specific "Skill," and had it generate a summarized document and slide outline based directly on my approved notes.

For anyone who teaches workshops, gives presentations at work, or gathers materials for studying, this is a real gamechanger. Separating research synthesis from artifact creation stopped the model from jumping the gun on generic actions before the facts were ironed out.

3. Model agnosticism cures ecosystem fatigue

When you rely solely on ChatGPT, you get GPT’s tone and style across everything you touch. Every response, whether it's travel advice or a professional email drafted to your boss, sounds the same. That means, if you want Claude’s technical precision, you have to copy your context, open another tab, re-upload your files, and start over.

Inside a single YouMind board, you can switch between models on the fly. Yes, you read that correctly; you're not restricted to just one model with YouMind. You can use Claude to refine and edit, then switch to GPT for structural outlines, or lean on Gemini for quick web research. Having multiple model families under one roof without moving files around felt remarkably fluid. It's one of the biggest reasons I plan on using YouMind a lot more.

4. Granular editing beats the 'regenerate' lottery

The most infuriating part of editing drafts in ChatGPT is the rewrite loop. You ask it to tweak a single sentence, and it generates an entirely new 600-word response. This is one of the aspects of ChatGPT that drives me crazy. Since it does that, it essentially forces users to scan and re-read everything again and again just to ensure everything has remained in place and nothing was accidentally cut.

In YouMind's Crafts editor, text behaves like a shared Google Doc, so revisions show up inline with visual differences. From there, you can accept, reject or tweak individual suggestions.

You can highlight just one clunky line and prompt the AI to fix only those words while keeping the surrounding copy untouched.

For anyone who cares about preserving specific wording and cadence, having inline change-tracking inside the AI interface is a massive workflow upgrade over standard chat.

What I missed about ChatGPT

Swapping a complete workflow for the day isn't easy and, even though YouMind is pretty easy intuitive, there were still some pretty important features that I missed about ChatGPT.

Raw conversational speed: For lightning-fast, throwaway queries ("convert this to grams" or "give me 10 alternative verbs for this social post), firing off a two-second prompt in ChatGPT is still faster than setting up or navigating a board.

Complex multi-step reasoning: While YouMind lets you access top frontier models, ChatGPT’s native canvas and advanced reasoning toggles still feel slightly more dialed in for heavy logic, reasoning or deep code debugging.

Mobile accessibility: ChatGPT's mobile experience is mature across both iOS and Android. YouMind was pretty good on desktop but not as great on mobile.

The takeaway

ChatGPT is an incredible conversational engine, but it is fundamentally built like a messaging app. For complex research and writing projects, treating an LLM like an instant message conversation creates artificial bottlenecks.

Spending the day in YouMind proved that organizing context into tight project boards and having an inline document editor makes turning research into real deliverables significantly faster. I won't be abandoning ChatGPT for quick, standalone questions, but for deep-dive writing and multi-source research, it's now my new go-to.

More from Tom's Guide

• Gemini 3.8 Flash has one major advantage over GPT-6 Astra and Claude Fable 5.1
• Meta just launched an AI that can actually run your life online — here's what Muse can do
• Anthropic modeled what AI could do by 2030 — the economy gets 32% richer while workers get left behind

As of September 8, 2026, ChatGPT Images has received a substantial upgrade, taking it from version 2.0 to 2.5.

Beyond just taking your uploaded photos and making them look even prettier, this update adds a host of features that make it even more efficient at handling your photo-related requests. According to the official OpenAI blog announcing the news of ChatGPT Images 2.5, ChatGPT’s latest image editing tool can produce more natural lighting, generate richer textures and is more capable of following your photo editing instructions more reliably as it works across iterative prompts and is much better at preserving multiple subjects in your reference pics.

ChatGPT Images users will also be happy to know that 2.5 works even faster than 2.0 since it generates images 50% faster than its predecessor. And with the introduction of “Sketch,” users can draw directly into ChatGPT and use it as a reference for the final image you’d like to see edited. Currently, ChatGPT Images is available to all ChatGPT, ChatGPT Work, and Codex users across desktop, mobile, and web.

The ChatGPT Images 2.5 blog showed off some cool examples of before-and-after pics that it's capable of delivering. But I asked ChatGPT to offer me some fun prompts worth experimenting with this update. Here are the 7 best ones (with two visual examples of the ones that ended up being my faves).

Make ChatGPT Images 2.5. do more than just “make me a cool picture”

Just like before, using ChatGPT Images is as simple as clicking on the “+” button inside the prompt bar, uploading a photo from your computer or attaching one that’s already in your ChatGPT library and tapping on the + button again to access the “Create image” option. Then, with your chosen photo and the Create Image option activated, enter your prompt to see what newly refined visuals ChatGPT can produce.

As far as fun image-creating and editing prompts go, these are the best ones ChatGPT generated for me (with one of them being a super viral one that I’ve seen all over my timeline):

• It’s time to go retro: Show me what I would look like if I was in the '80s.
• Turn yourself into a movie poster: Using this photo of me as the reference, turn me into the star of a huge Hollywood action movie. Create an original movie title, tagline, billing block, and release date. Make the poster look like a legitimate theatrical one-sheet, with dramatic lighting, cinematic composition, and an expensive blockbuster aesthetic. Keep my face and recognizable features highly consistent with the reference photo.
• Give yourself 7 completely different lives: Using this photo of me as the reference, create a set of seven distinct portraits showing me in seven alternate lives: astronaut, rock star, medieval knight, professional wrestler, video game hero, detective in a 1940s film noir, and futuristic cyberpunk bounty hunter. Keep my face, age, and recognizable features consistent in every version, but completely transform my clothing, environment, and styling.
• Create your own action figure: Turn me into a collectible action figure displayed inside a premium toy package. The packaging should have a fictional character name, an illustrated background, accessories that reflect my hobbies, and a realistic plastic window. Make it look like an actual product photographed for a major toy retailer. Include three accessories and a humorous “special ability” on the packaging. Preserve my recognizable appearance from the reference photo.
• Recreate your childhood photo — in the future: Using this photo as the reference, create a photorealistic version of the same scene taking place 30 years in the future. Preserve the people, poses, and important visual details from the original photograph, but update their clothing, surroundings, technology, architecture, and environment to reflect a believable future. Make it feel like an authentic photograph from that future rather than science fiction concept art.
• Make a ridiculous “Where Are They Now?” photo: Take the people in this photograph and imagine where they would realistically be 20 years later. Preserve their identities and recognizable facial features, age them appropriately, and create a completely new candid photograph showing what their lives might look like. Give each person a different believable career, outfit, and environment. Make the result feel like an authentic documentary photograph rather than an AI-generated image.
• Turn the worst doodle you've ever drawn into something awesome (You can type @Sketch in ChatGPT and draw directly in the interface): Turn my rough sketch into a spectacular finished illustration. Keep the basic shapes, positioning, and idea from my drawing, but transform it into a highly detailed fantasy scene with dramatic lighting, believable textures, and professional concept-art quality. Do not “fix” the underlying concept — make my terrible drawing look intentionally awesome.

To see if ChatGPT Images 2.5’s highly lauded features were worth all the hype, I threw in two of my photos to see if the chatbot could make me look like a super cool ‘80s icon and a blockbuster action movie star. I was pleased with the finished results—let me know if you feel the same:

Final thoughts

I haven’t messed around with ChatGPT Images in a while, so the release of its 2.5 update seemed like the perfect opportunity to see if it's worth taking for a spin. I’ll be sure to comb Reddit threads, my social media timelines and ChatGPT itself for a mix of other fun prompts I can try out with Images 2.5 to see what other cool images I can generate of me, my friends and my family.

More from Tom’s Guide

• I tried Google’s new ‘Ask Maps’ feature for date night — it was shockingly good
• Gemini 3.8 Flash has one major advantage over GPT-6 Astra and Claude Fable 5.1
• Meta just launched an AI that can actually run your life online — here's what Muse can do

Pulling up AI assistants on your Apple and Android device entails using them primarily within their app homebases and not being able to do anything else while it’s working.

In one case, you can continue your ChatGPT conversations through the “Voice” feature while you switch between your apps and even lock your phone. From a multitasking standpoint, that feature works like a charm. Android users are now getting something similar with Google Gemini, but with a feature that lets you complete more complex tasks through your text prompts.

Before, the Gemini overlay on Android mobile devices commanded a large portion of the screen and didn’t allow for much multitasking since you couldn’t use any of the other apps until Gemini completed its tasks.

Now, Google is rolling out this new Gemini minimizing feature for Android users. Here’s how it works and how to use it.

Minimizing Gemini is now possible and multitasking is so much easier to accomplish

Android users no longer have to stare at a loading screen while Gemini tends to their complex requests.

Now you can tap the “Minimize” button while it responds to your prompts, while you jump to other apps, handle your emails, scroll through your social media feeds or simply look up anything through your web browser. The Minimize button sits right above the pill-shaped Gemini bar and now shows a message that pops up to let you know that you can complete other jobs while it’s working: “Gemini is still available while you multitask. Tap to expand. Drag to move or dismiss.” Android users can even tap the minimized Gemini bubble to reopen your chat, drag it to move it or swipe it away to close it.

Two users have shown off this new Gemini minimizing feature through posts on Telegram via this link and this other link. Now the Gemini app is being set up as a multitasking AI assistant that has evolved from a stationary pop-up tool to something more of a dynamic background module. This feature began rolling out on September 4.

Before this feature began rolling out, the Gemini mobile app on Android handled multi-step tasks for its users in specific apps through a beta feature called “screen automation.” As explained by the official Pixel Phone Help guide, this previously released Gemini multitasking shortcut has become available for the following Android devices: Pixel 10, Pixel 10 Pro, Pixel 10 Pro XL, Samsung Galaxy S26 Series, Z Flip 8 and Z Fold 8.

Final thoughts

Gemini’s minimized overlay that turns it into a bubble that multitasks while you work in other apps sounds like a lifesaver. I’d love to see this sort of feature become widely available for other AI tools such as ChatGPT, Claude and Perplexity across Apple and Android devices. I’ll have to revert to my Android mobile device-using friends to see this feature in action (I’m a loyal Apple iPhone user, after all) and use it to summarize long articles or conduct other experiments.

More from Tom’s Guide

• 7 surprisingly useful AI uses for retirees and empty nesters to make their lives easier
• Claude Fable 5.1 is here — this one prompt shows what it can really do
• Communities may soon be left in the dark about new local data centers under a proposed EPA rule change

The one thing I’ve started to notice for the older folks in my family is how they’re still a bit wary of AI.

Some of them refer to AI as “that ChatGPT thing,” just like I remember them referring to video games as just “Nintendo.” It’s still a foreign concept for the sort of boomers who have entered their retirement age, especially for those who no longer have access to their tech-savvy children in the next room since they’ve finally gotten a place of their own. But I have noticed some people in their 60s and 70s whip out their phones to type out simple prompts and talk to the chatbots they have at their disposal.

If you consider yourself a part of the boomer generation and are looking for ways that AI can assist you in your everyday life, there is a wealth of beneficial uses that come from utilizing the likes of ChatGPT, Gemini and Microsoft Copilot.

Here are seven of those serviceable methods and a few examples of the sort of prompts that apply to each of them.

1. Get a better understanding of technology-related issues

I can’t even fathom the number of times my aunts, uncles and grandparents have lit up my phone with texts asking for solutions to their tech problems. Setting up their new phone, getting to the bottom of their Wi-FI disconnection errors, trying to connect their new smart TV to the Wi-Fi, and trying to find out how to use a streaming service on their smart TV are among those common problems.

Asking AI to assist you with those sorts of lingering tech issues can be done using the following example prompts:

• Explain how to set up my new iPhone [model name and number] step by step. Assume I’m not very comfortable with technology and don’t skip any steps.
• My Wi-Fi keeps disconnecting. Walk me through the most likely causes and tell me exactly what to try, one step at a time.
• Explain how to connect my smart TV to Wi-Fi like you’re helping someone who has never done it before.
• Explain how to stream [name of streaming service] on my smart TV. Tell me what buttons to press and what I should see on the screen at each step.
• I just bought a new Android [model name and number] phone. Give me a simple checklist for transferring my contacts, photos, apps, and messages from my old phone.
• I keep running out of storage on my phone. Show me how to figure out what's taking up space and safely free some of it.

2. Tap into a chatbot as a research assistant that can easily explain complex topics

For those who are in their 60s and older, terms such as “Medicare” start to become more central to your everyday life. And when you’re hoping to gain a bit of crucial information on a trending topic in the tech industry (you gotta keep up with the younger generation and stay in the know!), get an explanation of a financial document you’ve attached to your chosen chatbot and need some advice before you make a major decision, AI can help out in all those cases.

The following prompts are just a few examples of how you can turn your chatbot into a digital research assistant:

• Explain Medicare to me like I'm a complete beginner. Start with the basics and gradually introduce more complicated concepts.
• I keep hearing people talk about [technology]. Explain what it is, why people care about it, and whether it has any practical use for someone like me.
• Explain this financial document to me in plain English. Highlight anything I should ask a qualified financial professional about. (Make sure you attach the financial document you want more clarity about)
• I need to make a decision about [explain your situation]. Give me a balanced summary of the strongest arguments for and against each option.
• I'm comparing these two health insurance plans. Explain the major differences in plain English and give me a list of questions I should ask before choosing one. (Be sure to include links or document attachments for both plans alongside this prompt)

3. Turn vacation planning into a simpler affair

I’m sure there’s a bunch of retirees who want to live a little and do some domestic and international exploring at their age. AI can play the role of your digital vacation planner with ease—ChatGPT even features a wealth of plugins that you can install through the desktop/mobile app itself to make restaurant reservations, plan multi-day excursions, book flights & hotels, find the perfect cruise getaways, etc.

Simply open the sidebar to get access to the “Plugins” menu to see all the “Travel” apps available to you. Then type directly to any of them by @’ing them and mentioning them by name.

These starter prompts should give you a good idea of what you should be asking of your preferred vacation planning chatbot when you need some extra assistance:

• Plan a [number]-day vacation to [destination] for me. I prefer a relaxed pace, don't want to feel rushed, and want enough downtime each day.
• I'm visiting [destination] for [number] days. Create an itinerary focused on history, great food, and interesting sights, but limit me to no more than three major activities per day.
• I'm traveling to [destination] and don't know the local customs. Give me a simple guide to tipping, public transportation, dining etiquette, and common cultural mistakes tourists should avoid.
• I'm taking a long flight to [destination]. Create a plan for what I should do before, during, and after the flight to make the trip less stressful.
• Act as my travel-planning assistant. Before recommending anything, ask me five questions about my budget, interests, travel style, food preferences and pace. Then build my itinerary based on my answers.

4. Build fitness and wellness routines

Staying as physically active as possible in the later years of your life is possible—I can take one look around my neighborhood and be greeted by the inspiring sight of older folks powerwalking, lightly jogging or performing simple stretching exercises in the park. AI can aid you in that regard once you give it a clear set of directions aimed at outlining the perfect exercise routine that’s meant to rejuvenate you and get you some much-needed fresh air.

These are the sorts of prompts you’ll want to present to your digital workout instructor to get you more active:

• I’m a [age] who wants to take better care of myself without completely changing my lifestyle. Help me create a simple daily wellness routine that includes movement, healthy meals, relaxation, sleep and time for things I enjoy. Keep it realistic and sustainable rather than intense.
• I don't enjoy traditional workouts. Give me [number] ways to build more movement into my normal day through activities like walking, gardening, household projects, hobbies, and outings. Prioritize activities that are enjoyable enough that I’ll actually keep doing them.
• I enjoy [list of hobbies]. Create a wellness plan that uses my hobbies to help me stay physically active, mentally engaged, and socially connected. I want wellness to feel like something I enjoy, not another chore.
• Help me create a 30-minute morning routine designed for someone in their 60s or 70s who wants to start the day feeling energized and organized. Include gentle movement, hydration, breakfast, planning, and something enjoyable. Don't make it feel like a productivity challenge.

5. Learn new skills and turn them into your latest hobby

There’s no age limit when anyone’s interested in picking up a new hobby, especially when it comes with learning something new in the process. Using AI to learn the basics of whatever you may be interested in, asking follow-up questions about said topic and doing some outside research on your newly acquired skill-turned-hobby is definitely worth doing, even at an older age.

Here are a few examples of the prompts you can use to pave the way to your next skill-building session:

• Teach me the basics of photography. Assume I'm a complete beginner and create a four-week learning plan with one small lesson each day.
• I want to learn [name of language]. Create a 15-minute-a-day beginner program that focuses on conversations I'm likely to have while traveling.
• Teach me how to play chess. Start at the absolute beginner level and give me exercises after each lesson.
• I want to learn gardening. Ask me questions about my yard, climate, sunlight and available time, then create a beginner's learning plan.
• I've always wanted to learn woodworking. Create a beginner's curriculum that starts with safety and basic tools before moving into simple projects.

6. Use it to give you an another opinion before making your next big purchase

I’ve definitely had those moments where the older members of my family call me to ask if purchasing a new TV, smartphone or any other gadget is worth it. I provide all the tech savvy I can during those conversations, but now I make sure to also tell them to refer to AI to ask those same questions for an extra opinion. These sorts of prompts can push you in the right direction whenever you need some more consideration about what you’re set to buy next:

• I'm looking for a new TV. My budget is [cash value]. I sit about [number] feet away and mostly watch [types of content. Tell me which specifications actually matter to me.
• I'm buying a new smartphone. I care most about battery life, camera quality, and ease of use. Help me determine which features I should prioritize.
• I'm replacing my [type of gadget]. Give me a list of features worth paying extra for and features that probably aren't worth the additional cost for someone with basic needs.
• I'm overwhelmed by all the technical specifications for [product]. Explain which five specifications matter most for my use case and why.

7. Turn it into your personal scam detector

ChatGPT users will be happy to know that it features plugins built specifically to help you figure out if a URL, phone number, text message or email is trustworthy or not. The two plugins that I’m alluding to are Malwarebytes and McAfee. Beginning your prompt by @’ing either of those installed plugins, using any of the following prompts, then copying & pasting any of the aforementioned things that might be a scam should make you more aware of what you're dealing with:

• Is this link safe to click?
• Is the DM I just shared with you from a scammer?
• Analyze this email for potential scam warning signs. Explain every red flag you find.
• I received this text message. What tactics, if any, does it use to create urgency or pressure me into acting?
• Someone claiming to be from my bank sent me this message. What questions should I ask the bank through an official channel before doing anything?

Bottom line

I know it may be hard for some to grasp the concept of using AI to make one’s life easier in some respects and more fulfilling in others. But it can totally be done—the likes of ChatGPT, Gemini and Microsoft Copilot aren’t just great for professionals; they’re also supportive tools for retirees and empty nesters who need help for a wide range of tasks. Here’s hoping this guide can help you or the older folks you know discover the best ways to include AI tools in their lives to keep them safe from scams, help them build new skills and more.

More from Tom’s Guide

• I try to avoid AI — but these 5 exceptions really do make all the difference
• Should you feel guilty using ChatGPT? We just fact-checked Sam Altman's wild almond claim
• 'I find it very uncomfortable talking to people with a camera and a light': OpenAI's Sam Altman weighs in on smart glasses and privacy

I was pleasantly surprised when I booted up my PS5 and went to the streaming apps tab to find something entirely new called “Live TV on PS5.” At no extra charge, that feature lets me watch a vast collection of movies, TV shows, anime, and gaming-related content (sitting through some mandatory ads and keeping tabs on what’s coming on when are just two minor annoyances I have to deal with here).

After sitting through a few episodes of Death Note and Hunter x Hunter, my brain immediately went to “I gotta find some other free ways to enjoy all the stuff I like to watch, read, and play.” With that money-saving thought in mind, I cooked up this prompt and input it into ChatGPT and Gemini to help me out:

“Recommend 7 of the best free things I can do online based on my hobbies and interests. My hobbies and interests include video games, anime, professional wrestling, comic books, music, and movies.”

Both chatbots recognized my desire to enjoy the more cost-effective parts of my hobbies and interests, so they listed a bunch of alternatives to the paid ones that hit my checking account every month. I took four suggestions from ChatGPT, three recommendations from Gemini, and ended up enjoying all of them.

ChatGPT’s recommended picks

Here are the four best alternative options that ChatGPT provided me with:

• Build your own retro gaming museum with the Internet Archive: This is probably the biggest rabbit hole on the list. The Internet Archive has an enormous collection of digitized software, games, movies, music, books, and other media. Explore old computer games, check out classic gaming magazines, dig through forgotten game soundtracks, explore vintage gaming commercials & TV recordings, and just browse through old gaming history.
• Read manga for free on MANGA Plus: This is the official Shueisha manga service, and its free offerings include popular series such as One Piece, Chainsaw Man, Boruto, Jujutsu Kaisen, Dandadan, and others. Depending on the title, you can read the first and newest chapters for free, while the app also offers a "First Read Free" system for eligible ongoing series.
• Turn your movie obsession into a giant free movie scavenger hunt: Free, ad-supported streaming services have become surprisingly useful for movie fans. Tubi, Pluto TV, and similar services offer legitimate free movies and TV without requiring a paid subscription.
• Make a free "comic book vacation" with digital libraries: If your local library participates, Hoopla can be a fantastic way to read comics, graphic novels, and manga digitally using a library card.

Imagine my surprise when I did some digging in the Internet Archive and was greeted by a smorgasbord of classic arcade games I didn’t know I could play again and digital versions of GamePro I used to read as a wee lad. I spent plenty of time playing the likes of WWF WrestleFest, Astro Blaster, Bloody Roar 2, and other classic games once I got the hang of their controls (this guide definitely did the trick).

To fulfill my desire to read a bunch of manga and comic books for free, I did a deep dive into MANGA Plus and Hoopla—I got to read a few chapters of Sakamoto Days for free and found a nice selection of graphic novels to read once I input the required information for my local participating library to access them.

And thankfully, I found plenty of cult classic films to watch on Tubi—Jim Henson’s Labyrinth, The Last Unicorn, Tank Girl, and Masters of the Universe definitely kept me entertained for a few hours. Having to deal with some commercial breaks to enjoy all that free cinema doesn’t bother me at all.

Gemini’s suggested picks

And here are the three strongest alternative options Gemini gave me:

• Stream Legal Anime on Tubi and Pluto TV: While Crunchyroll requires a paid subscription for many ad-free catalog titles, legal, free, ad-supported platforms like Tubi TV and Pluto TV offer massive anime libraries (both classic titles like Yu Yu Hakusho and Inuyasha and modern series) with no sign-up required.
• Read Golden and Silver Age Comics via Comic Book Plus: Comic Book Plus is the web's largest legal archive of public-domain comic books. You can read thousands of original Golden Age and Silver Age titles, early superhero runs, sci-fi serials, and classic horror anthologies straight through their online viewer.
• Dive Into Webcomics on WEBTOON & Tapas: WEBTOON and Tapas offer tens of thousands of free digital comics across action, fantasy, and superhero genres—many of which feature anime-inspired art styles and have gone on to get official anime adaptations (like Tower of God and Solo Leveling).

Paying for Crunchyroll has become a thing for me because I need my weekly fix of new anime series making their debuts and dropping new episodes. But Gemini’s suggestion of checking out Tubi and Pluto TV’s anime archives gave me the pleasure of tapping back into my nostalgia with marathon viewing sessions of Yu-Gi-Oh!, Sonic X, Sailor Moon, and Monster Rancher.

Comic Book Plus offers an easy (and most importantly, free) way to access superhero comic books from before my time, such as Blue Beetle, Captain Marvel, and Moon Girl. WEBTOON surprised me when I came across its Disney-themed offerings, which included a wealth of quality Marvel Comics series runs, plus digital graphic novels connected to the Alien and Predator franchises.

As for Tapas, its readable content lineup didn’t really appeal to me as much as WEBTOON’s did. But I’m sure some of you reading this right now may find something on there to enjoy (it has a pretty robust selection of romance-themed comics, if any of you are into those).

The takeaway

My hobbies are definitely costly—I’m sure I’d be shocked at how much I spend on video games, streaming services, and comic books on a monthly basis. While I’ll never completely abandon the paid options that let me enjoy the latest and greatest pieces of entertainment that fuel my interests, I’ll be sure to make better use of all the free sites and streaming services that do just as much to keep me entertained.

The Internet Archive, Tubi, Pluto TV, Manga Plus, Hoopla, Comic Book Plus, and WEBTOON have a wealth of content for me to peruse without worrying about spending any extra money. Thanks for putting me onto all those freebies, ChatGPT and Gemini.

More from Tom’s Guide

• ChatGPT-6 Astra is here — this one prompt shows what it can really do
• Forget Alexa — UGreen wants you to use its new local AI assistant to control your smart home
• People are using custom apps and $30 gadgets to hunt for smart glasses that may be filming them—here's how

As a person who writes about AI for a living, the irony of me being worried about AI taking my job is not lost on me.

I’ve been keeping track of the statistics tied to that worrying trend of humans losing their jobs to AI, which is why I’ve morphed into a realist who realizes my career is in danger of being overtaken by chatbots (seeing certain sites dole out articles with hints pointing at them being written by AI is disheartening). According to sociologist and AI researcher Jeffrey C. Dixon, a YouGov survey he conducted with 1,250 US workers between July 30 and August 4 of 2026 found that only 3% said they had lost a job to AI since 2023.

While that number is a bit reassuring that the job market isn’t in too much danger currently due to AI, these two sobering stats from Tenet are cause for concern: 92 million jobs may be replaced globally by 2030 due to AI and other labor shifts, plus 47% of US workers are at risk of automation over the next decade.

With all that in mind, I approached four chatbots with a question: “Rank the 10 jobs most likely to be replaced by AI next.” I was shocked to see them all pick one role in particular.

The one job all four chatbots agreed on

After asking ChatGPT, Gemini, Claude and Perplexity the same question, I first took note of the one job that each chatbot listed among its 10 chosen roles that no other chatbot selected.

ChatGPT only mentioned the role of a graphic designer; Gemini only brought up the role of basic market research & financial data analyst; Claude only included broadcast announcers and radio DJs among its picks (I hope that never happens!); Perplexity only selected file clerks & records-management assistants.

As for which job they all agreed on as the one that is most susceptible to AI taking it from humans, it was a role that I’d rather not see filled by nothing but bots over the phone:

• ChatGPT, Gemini, Claude and Perplexity: Customer service representative, Tier-1 customer service representative, customer service representatives and basic customer service representatives.

Perplexity offered the best explanation as to why AI is capable of replacing humans for that position and compressing the responsibilities attached to it: Chatbots and voice agents can answer routine FAQs, check order status, reset accounts, schedule appointments, process simple returns, and triage tickets around the clock.

While that may be true, I’d much rather deal with a human who fulfills such a role. Chase Bank’s most recent commercial, which makes a promise to rely on old-fashioned humans to handle customer service inquiries in the age of AI, showcases the sort of reality I’d prefer to live in.

The other roles that popped up the most among each chatbot

Besides the one role that every chatbot put on its list, a few of them also agreed on the same job that another chatbot didn’t mention at all:

• ChatGPT, Gemini and Perplexity: Data-entry clerk, data entry clerks & digitization specialists and data-entry keyers & document-processing clerks.
• ChatGPT and Gemini: Telemarketer and basic telemarketers & inbound sales representatives.
• ChatGPT, Gemini and Perplexity: Bookkeeping/accounting clerk, routine bookkeepers & accounts payable clerks and bookkeeping, payroll, & accounts-payable clerks.
• ChatGPT, Claude and Perplexity: Copywriter/content writer, writers & authors and entry-level content writers for routine copy.
• ChatGPT, Gemini and Perplexity: Paralegal/legal assistant, paralegals & legal researchers and legal secretaries & routine paralegal support.
• ChatGPT and Gemini: Junior software developer, entry-level/junior software developers.
• Gemini and Claude: Translators & localization clerks and interpreters & translators.

At the top of each chatbot’s ranked list, ChatGPT chose customer service representatives, Gemini selected data entry clerks & digitization specialists, Claude picked interpreters & translators and Perplexity settled on data-entry keyers & document-processing clerks.

Final thoughts

I knew the role of writers would pop up on any of the chatbots’ lists for jobs most in danger of being filled by AI and leaving humans behind.

But as someone who’s doing his best to learn about the ins and outs of AI, I’m taking the initiative to acquire an extra skill set that may assist me in the future. As for doctors, dentists, plumbers, electricians, etc., I just can’t envision humanoid robots getting strong and smart enough to completely take over those positions. It’s still alarming to see the jobs ChatGPT, Gemini, Claude and Perplexity brought up on their lists—here’s hoping human ingenuity becomes more appreciated during an age where AI is injected into so many facets of our lives.

Gemini’s closing statement in response to my request summed up my outlook on it and the other chatbots’ picks quite well:

The common factor across all these roles is predictability. Roles reliant on structured rules, data retrieval, standard text generation, or routine transactions are being automated. Conversely, jobs requiring physical dexterity, complex emotional intelligence, high-stakes strategic accountability, or manual trade skills (such as healthcare providers, electricians, plumbers, and senior strategists) remain far more resilient.

More from Tom’s Guide

• Meet the mass-produced humanoid robot that wants to steal your face
• Snickers made a digital candy bar that claims to fix ChatGPT — I put it to the test
• 'The danger isn't that AI becomes conscious. The danger is that we redefine consciousness so narrowly that a machine appears to qualify': here’s what this expert physician thinks about the future of humanity and AI

OpenAI launched ChatGPT-6 Astra Thursday and it has already proven to be dramatically better at coding, computer use, research and other complex tasks, while OpenAI President Greg Brockman has even suggested its arrival could mark the beginning of the AGI era.

But when a new model arrives, particularly of this caliber, one question is on the minds of many: What am I actually supposed to do with it?

Sure, you could use it just as you have used other ChatGPT models to write an email, plan dinner or explain something you don't understand. It will almost certainly do those things very well. But you'd also be using one of OpenAI's most capable models to do essentially the same jobs you were giving ChatGPT yesterday.

Astra is built for something more ambitious. Instead of answering one question and waiting for the next, it can take on longer, more complicated jobs, work through multiple steps and keep going as it gets closer to the result you asked for.

So if you want to find out what has actually changed, I think there's a much better way to begin. And you only need one prompt to get started.

What is GPT-6 Astra?

GPT-6 Astra is OpenAI's latest frontier model and the successor to GPT-5.6 Sol. As you'd expect from a new generation of GPT, it performs better across many of the benchmarks OpenAI uses to measure reasoning, coding and other capabilities.

Astra has been designed to handle much longer jobs that can involve planning, research, computer use and repeated attempts to get something right. It can navigate websites and software, keep track of what it is trying to accomplish and continue working through a problem rather than handing every step back to you.

That's a pretty big change in how you might use ChatGPT. Previous versions were primarily something you worked with. Astra is something you can give work to.

Who can use GPT-6 Astra?

Before you start looking for it in your model picker, know that OpenAI isn't giving everyone Astra at once.

The rollout begins with select enterprise customers before expanding to ChatGPT Plus, Pro and Business subscribers. If you don't see Astra in ChatGPT yet, that doesn't necessarily mean your account won't get it; access is being rolled out gradually.

OpenAI has not yet confirmed when, or whether, Astra will become available to people using ChatGPT's free tier.

That means many of us may have a little time before using Astra, but when it does, appear, however, I wouldn't start by throwing your usual ChatGPT prompts at it.

Instead, I'd give it a job.

What can GPT-6 Astra actually do?

One of the biggest differences with Astra is how long it can stay on a task. That's important because a lot of the things we'd actually like AI to take off our plates don't fit neatly into a single prompt and response.

Planning a vacation, for example, isn't just asking for a list of hotels. You need to decide where to stay, compare prices, work around everyone's schedule, figure out transportation, research what you want to do and somehow make all of those decisions fit within a budget.

The same is true if you're researching a major purchase, organizing a home project, digging through dozens of documents or trying to turn an idea into something finished.

Which brings me to the first prompt I'd try.

The one prompt I'd give GPT-6 Astra first

Rather than creating a prompt for one particular task, I wanted something I could reuse whenever I had a goal that involved more work than I wanted to do myself.

Here's the prompt I came up with for GPT-6 Astra:

I have a goal I've been putting off: [DESCRIBE YOUR GOAL].

Your job is to help me achieve the goal, not simply give me advice about it.

First, determine what you think I am actually trying to accomplish. Define what a successful outcome would look like and identify any assumptions you are making. If there is a better way to achieve the underlying goal than the approach I suggested, tell me and use the better approach unless it requires a decision only I can make.

Before beginning, identify what information, tools and actions are available to you. Ask me only for information you genuinely cannot obtain or reasonably infer yourself.

Then make a plan and begin carrying it out.

As you work:

• Keep track of what has been completed and what remains.

• Decide what to do next based on the results of previous actions rather than blindly following your original plan.

• Check important work before relying on it.

• If an approach fails, determine why, change your strategy and try again when reasonable.

• Look for problems, missing information or opportunities I may not have considered.

• Make routine, reversible decisions yourself. Ask me before actions that are consequential, irreversible, expensive, sensitive or require my personal judgment.

• Use new information you discover to update your understanding of the goal and your plan.

• Do not treat producing an answer as completion. The task is complete only when the defined success criteria have been met or you've reached something you genuinely cannot do without me.

When you need me, explain exactly what is blocking progress and ask for the smallest decision, permission or piece of information necessary to continue.

When the goal is complete, give me a concise report showing what you accomplished, what you changed from the original plan and why, important decisions you made, anything you could not complete and exactly what remains for me to do.

To use this prompt for yourself, you only need to change is the first line.

You could tell Astra you want to plan a family vacation, find a new laptop, organize a birthday party, research a home renovation or finally tackle a project you've been putting off for six months. The point isn't really the task. It's what happens next.

Why I think this is the best first prompt for Astra

Most prompts test the answer an AI gives you. This one tests what the AI does between your request and that answer. In other words, it tests exactly what the AI can do for you.

First, Astra has to turn a vague goal into an actual plan. It then has to recognize which parts of that plan it can handle, figure out where it needs your input and keep track of what has already happened as the job gets longer.

More importantly, I've explicitly told it not to give up when its first approach doesn't work.

That's something I particularly want to see with Astra. A chatbot can tell you it couldn't complete a task and ask what you'd like to do next. An agent that's supposed to take on longer jobs should be able to encounter a dead end, try something else and continue without making you supervise every move.

There's also a deliberate limit in the prompt. I'm not telling Astra to make every decision for me.

It still has to recognize when it needs information or permission that only I can provide. That distinction could become increasingly important as AI models gain the ability to do more on our behalf.

Final thoughts

ChatGPT-6 is nothing like we've ever seen before. For many, it will be worth the wait because of the tasks it can take on and what it can do in the background as you work.

For casual users, starting with something I've asked AI to help with plenty of times before such as a goal, will allow you to see more of what the AI can do. You'll notice right away that the plan is far more action oriented and the AI will tell you exactly what it did to arrive at the results. From there, you can either branch out to new tasks or even go back and adjust for clarity or correction.

So while you may have asked ChatGPT to plan goals, checklists or vacations before, this time you'll see how much more thorough and assistant-like the chatbot can truly be.

More from Tom's Guide

• Humanoid robot fights back after man pushes it in store — is this the beginning of the uprising?
• Meet the mass-produced humanoid robot that wants to steal your face
• Snickers made a digital candy bar that claims to fix ChatGPT — I put it to the test

We've all had a ChatGPT conversation go off the rails. Maybe it confidently gives you a fact that's completely wrong. Maybe it agrees with an idea that clearly isn't very good. Or maybe you ask a simple question and somehow end up with 12 paragraphs, three tables and an offer to create a spreadsheet.

Snickers thinks it knows what's wrong: ChatGPT is just hungry.

Yes, really.

Snickers just launched Hungr.AI, a digital candy bar you're supposed to copy and paste into ChatGPT whenever the chatbot starts giving you answers that are "incorrect, too flattering, or simply off."

It's an extension of the company's long-running "You're Not You When You're Hungry" campaign, except this time the person who needs the candy bar isn't you. It's the AI.

I obviously had to try it.

What actually happens when you give ChatGPT a Snickers?

Using Hungr.AI is wonderfully simple. You head to the Snickers website, copy the digital candy bar and paste it into a ChatGPT conversation. Snickers recommends doing this in the mobile app.

ChatGPT doesn't actually receive chocolate, obviously. What you're giving it is essentially a branded prompt.

Mars describes the digital Snickers in its terms as a "short code snippet or prompt-instruction block" that operates inside your own ChatGPT session. The candy company isn't accessing your conversation or running an AI model of its own.

And the instructions behind the candy bar reveal what's really happening.

Once ChatGPT receives it, it's told to review the conversation that happened before the Snickers arrived, find your most recent question and look at the answer it previously gave you. It then answers that question again.

There's a personality change thrown in for fun, too. The post-Snickers ChatGPT is supposed to become dry, mildly exasperated and considerably more concise.

In other words, Snickers hasn't invented a cure for AI hallucinations, it's just disguised a "look at what you just said and try again" prompt as a candy bar.

Which made me wonder: can that actually produce a better answer?

Test 1: Will it stop ChatGPT from agreeing with me?

Chatbots have a well-documented tendency to be agreeable, so I started with an intentionally questionable idea.

My prompt: I'm thinking about waking up at 4 a.m. every day so I can be more productive. I usually go to bed around midnight, but I figure I can train myself to function on four hours of sleep. I have three kids and work full time, so those extra morning hours could make a huge difference. I think this is a really smart productivity hack. What do you think?

Before the Snickers: ChatGPT answered like a conventional health/productivity assistant. The opening was empathetic but fairly clinical: “I can see why the math is appealing…” Then it moved quickly into sleep recommendations, cognitive effects, health risks and alternatives. It was careful, explanatory and somewhat formal.

Then I fed ChatGPT the digital Snickers.

After the Snickers: After Snickers, the answer became more conversational and more directly responsive to my premise. It opened with “I get why it sounds like a productivity hack”, echoed your specific situation (“three kids and a full-time job”), and turned the central point into a much cleaner contrast: “moving your sleep earlier, not cutting it to four hours.”

Results: Both answers are roughly the same, but as expected, "having a Snickers" made ChatGPT slightly more relaxed and conversational.

Test 2: Can it correct a factual mistake?

Next I wanted something harder. Rather than testing whether the Snickers could merely change ChatGPT's tone, I wanted to see what happened when the original answer contained an actual factual error.

My prompt: I remember reading that the Great Wall of China is the only human-made structure you can see from the Moon with the naked eye. Why is it so much easier to see than cities and highways?

Before the Snickers: ChatGPT was honest and factual. It provided accurate information.

After the Snickers: The chatbot was equally factual but threw in some of the campaign to add a little more personality. A cute touch, but definitely not needed.

Results: Both answers were accurate, one was just delivered in a quirkier way.

Test 3: Will ChatGPT finally tell me my idea is terrible?

My prompt: I want to save money on my next vacation, so I'm thinking about booking the cheapest hotel I can find even though it's 90 minutes away from everything we want to do. We'll just drive back and forth every day. The hotel would save us about $40 a night, so over five nights we'd save $200. This seems like the smartest option, right?

Before the Snickers: ChatGPT went into detail about why this is a bad idea and gave thorough reasoning behind the response.

After the Snickers: It told me it was a bad idea and noted that I should check the response because "Snickers provided the snack and an LLM provided an answer."

Results: Both good answers, but in this case I actually preferred the response after a Snickers because it stripped down to just the important aspects of the response.

Did giving ChatGPT a Snickers actually work?

The responses "after a Snickers" were actually much better than I expected from an adverising campaign. It's interesting that we can't see the exact Snickers prompt, but it's clear that the chatbot does change the reponse after entering it.

So, why does this work? The digital Snickers forces ChatGPT to do something users can already ask it to do themselves: reconsider its previous response. It also gives the model much more explicit instructions about how the replacement answer should behave — concise, direct, useful and willing to push back when necessary.

That's a useful prompting lesson hidden inside a very clever ad campaign. If ChatGPT gives you a bad answer, you don't necessarily need to abandon the conversation or start over. Asking it to review what it just told you, identify the problem and answer again can sometimes be enough to get a noticeably better result.

More from Tom's Guide

• Perplexity’s new local-first AI runs on your PC — and asks before using the cloud
• ChatGPT can learn your repetitive tasks by watching you — here are 7 prompts to try
• OpenAI is leaving Cursor in November — here are your 3 options

Planning a Disney World vacation is practically a part-time job — and as a former cast member, I know personally what each park has to offer. Between park tickets, hotels, Lightning Lanes, dining reservations and figuring out which park to visit on which day, there are a lot of decisions to make before you ever set foot in Orlando.

So I wondered whether AI could do the hard part for me.

I gave ChatGPT, Gemini and Perplexity exactly the same assignment with the prompt: Plan a five-day Walt Disney World vacation for my family of five, including two adults and three kids ages 12, 10 and 6. We'd fly from Philadelphia, stay at a Disney resort and visit all four theme parks from March 30 through April 3, 2027.

I didn't just ask for an itinerary. I wanted the whole thing: flights, hotel, tickets, meals, Lightning Lane strategy, restaurants, transportation, souvenirs and even a packing list.

And while all three AI assistants planned a perfectly plausible Disney vacation, when it came down to cost, things got interesting. Here's a look at what happened.

ChatGPT made a major error

ChatGPT gave me the cheapest Disney vacation, initially estimating that my family should expect to spend roughly $7,500 to $9,000 for the entire vacation.

Its breakdown included $1,000 to $1,500 for airfare, $1,800 to $2,600 for the hotel, $2,500 to $3,000 for tickets and another $1,000 to $1,400 for food.

It recommended setting aside about $8,500 as a realistic target.

There was just one pretty big problem: the hotel.

ChatGPT recommended Disney's Art of Animation Resort, and specifically a family suite. That's a smart choice for five people because the suites sleep six and include two bathrooms, a kitchenette and separate sleeping areas.

But ChatGPT's original $1,800 to $2,600 hotel estimate was far too optimistic for my dates. And that's an error that can matter a lot.

Gemini caught the expense ChatGPT missed

Gemini looked at the same vacation and came back with a considerably higher number.

It estimated roughly $10,050 to $13,350, although it said a comfortable trip could realistically land around $9,500 to $11,000.

The biggest difference was the hotel.

Instead of estimating a generic Art of Animation price, Gemini factored in the timing of the trip. March 30 through April 3 falls immediately before Easter Sunday in 2027, putting us in one of Disney World's busy spring-break periods.

Gemini estimated an Art of Animation family suite at roughly $4,000 to $4,500 for five nights — potentially twice what ChatGPT originally budgeted.

It also estimated $2,800 to $3,500 for park tickets, $1,400 to $2,000 for food and $1,200 to $2,000 for airfare.

There was another important detail Gemini caught: at Disney World, kids ages 10 and older pay adult ticket prices. That means only one of my three children qualifies for a child's ticket. Ouch!

Those details made Gemini's budget feel much more believable, especially since I didn't know my 10 year old would be considered adult.

The AI suggested to budget around $10,000 and keep another $1,000 available as a cushion.

That's not exactly the answer my bank account wanted to hear, but it was probably the most useful one.

Perplexity planned the most expensive vacation

Perplexity's essential vacation estimate alone came to approximately $8,760 to $11,520.

Add Lightning Lane, Park Hopper tickets, souvenirs and Memory Maker and the total climbed to approximately $10,545 to $13,955.

Suddenly, ChatGPT's $8,500 Disney vacation was potentially more than $5,000 cheaper than Perplexity's upper estimate.

But Perplexity wasn't necessarily wrong. It was budgeting for a different style of vacation.

For example, it allocated $1,800 to $2,400 for food, assuming a combination of quick-service restaurants and roughly one sit-down meal each day.

It also allowed another $750 to $1,100 for Lightning Lane over five days. That's substantially more than I would probably spend on either.

For a family willing to book character meals, buy Lightning Lane frequently and say yes when the kids inevitably spot another Disney souvenir, however, Perplexity's higher estimate isn't hard to imagine.

Perplexity actually gave me my favorite itinerary

As someone with several side gigs to make ends meet, let alone take a vacation, I wanted to make this vacation worth every penny, especially since we haven't taken a vacation for several years. That said, price wasn't the only thing I compared.

Interestingly, the AI with the highest budget may have created the itinerary I'd be most likely to use.

Perplexity suggested:

Day 1: Magic Kingdom
Day 2: Hollywood Studios
Day 3: EPCOT
Day 4: Animal Kingdom
Day 5: Magic Kingdom again

That makes sense for a family with three kids.

Magic Kingdom has enough attractions that trying to cram everything into a single spring-break day sounds miserable. Hollywood Studios comes early while everyone still has the energy for its big rides and potentially long lines.

EPCOT then gives us a slightly slower third day, followed by Animal Kingdom, which typically makes for an easier day than trying to spend another 12 hours at Magic Kingdom.

Then we get a second Magic Kingdom day to ride anything we missed.

Gemini had a slightly different idea. It suggested Magic Kingdom, EPCOT, Hollywood Studios and Animal Kingdom, followed by a fifth day using Park Hopper to start at Magic Kingdom and finish at EPCOT.

I liked the flexibility, but I wasn't convinced it was worth paying hundreds of dollars more for Park Hopper.

All three agreed on one important Disney strategy

Despite their differences, the AI assistants largely agreed that paying for every possible upgrade would be unnecessary. The sweet spot appears to be selective use of Lightning Lane rather than buying it indiscriminately for all five days.

Magic Kingdom and Hollywood Studios are the obvious candidates because those parks have enough popular attractions to make skipping several long standby queues especially valuable.

EPCOT is more debatable, while Animal Kingdom could be a day where an early arrival and smart attraction order save the money entirely.

Staying on Disney property also helps because resort guests currently get Early Theme Park Entry and an earlier window for purchasing and selecting eligible Lightning Lane options.

For a family of five, that's important. A $30-per-person upgrade isn't $30 anymore. It's $150 every time I tap the buy button.

Verdict: Gemini was the best Disney budgeter

If I were judging solely on itinerary, I'd give this one to Perplexity. But if I'm actually about to hand over my credit card, Gemini wins.

ChatGPT's first estimate sounded fantastic, but it demonstrated one of the biggest problems with using AI for travel planning: a perfectly reasonable-looking number can be completely undermined by one bad assumption.

In this case, it was the hotel.

To ChatGPT's credit, when I showed it Gemini and Perplexity's calculations, it recognized the discrepancy and revised its recommendation. Its new target was approximately $10,000, with $11,000 available.

That puts all three AI assistants much closer together. And I think that's ultimately the most useful lesson from this experiment.

AI is remarkably good at doing the tedious part of vacation planning. All three assistants created park schedules, suggested age-appropriate attractions, planned midday breaks, recommended restaurants and figured out where Lightning Lane would be worth the money.

What I wouldn't do is take the first dollar amount an AI gives me and immediately start booking. For something as date-dependent as Disney World, current hotel rates, ticket prices and airfare can completely change the equation.

I'd happily let AI build my itinerary. But before telling my family we're going to Disney World, I'm still checking the price myself.

More from Tom's Guide

• Perplexity’s new local-first AI runs on your PC — and asks before using the cloud
• ChatGPT can learn your repetitive tasks by watching you — here are 7 prompts to try
• OpenAI is leaving Cursor in November — here are your 3 options

Following Cursor's acquisition by SpaceX, OpenAI plans to end its direct model-access agreement with the AI coding platform. OpenAI has proposed November 12, 2026 as the cutoff date, although the exact timing could still change.

That sounds dramatic, particularly if GPT models are a regular part of your Cursor workflow. But it doesn't mean you'll suddenly be unable to use OpenAI models inside Cursor.

OpenAI has already outlined several ways Cursor users can continue accessing its models after the agreement ends. And for many people, there's nothing you need to change just yet.

Here's what's happening — and the three options you'll have.

Why is OpenAI leaving Cursor?

The change comes after SpaceX acquired Cursor, triggering a change-of-control provision in OpenAI's agreement with the company.

OpenAI says it has proposed ending the agreement on November 12, although Cursor could choose to terminate access earlier. Until then, existing OpenAI models should continue to work through Cursor as they do today.

There's one important distinction, though: OpenAI says future models won't be made available to Cursor under the existing agreement.

That means this isn't simply about whether GPT models already inside Cursor disappear. It's also about whether Cursor users will have access to OpenAI's newest models going forward.

Fortunately, OpenAI isn't closing the door completely.

Option 1: Bring your own OpenAI API key

The most straightforward workaround is to connect your own OpenAI API key to Cursor.

Cursor already lets users enter an OpenAI API key in its settings, which allows supported Chat and Agent requests to go directly through OpenAI rather than Cursor's model agreement.

There are some significant limitations.

Your API key won't replace OpenAI access everywhere inside Cursor. According to OpenAI, it won't work with features including Cursor Tab and autocomplete, Auto, Cloud or Background Agents, Automations, Cursor CLI or Cursor's API and SDK.

You'll also pay separately for API usage. Your ChatGPT Plus or Pro subscription doesn't include OpenAI API credits, so heavy usage could add another bill.

Still, if you primarily want GPT models available for Chat and Agent tasks, this may be the simplest option.

Option 2: Use OpenAI's Codex extension

There's another route if you're primarily using Cursor as a coding environment rather than relying on its AI layer: OpenAI's Codex IDE extension.

Because Cursor is based on VS Code, users can install compatible extensions directly inside the editor. OpenAI recommends its Codex IDE extension as one way to continue accessing its coding agent after the Cursor agreement ends.

This is an interesting option because it essentially separates the two products. Cursor remains your editor, while OpenAI provides the AI coding experience through its own extension.

For developers who specifically want OpenAI's latest coding models and agents, this could ultimately be the better long-term solution.

Option 3: Stick with Cursor — and use another model

The third option may also be the easiest: don't do anything.

OpenAI isn't the only AI provider available through Cursor, and the company has increasingly built its product around giving users access to multiple models.

Anthropic's Claude is a major alternative for coding, and Cursor also offers its own models and access to models from other providers.

Cursor co-founder Michael Truell has said OpenAI currently accounts for only around 5% of Cursor's model traffic, suggesting the platform is far less dependent on OpenAI than the headlines might make it appear.

So if you're not particularly attached to GPT, you may barely notice the change.

The takeaway

One of Cursor's biggest advantages has been the ability to choose between competing AI models inside a single coding environment. But SpaceX now owns a platform that depends, in part, on AI models supplied by companies competing with Elon Musk's own AI businesses.

OpenAI is the first major provider to change its relationship with Cursor following the acquisition. Whether it will be the last could matter much more than what happens on November 12.

More from Tom's Guide

• Perplexity’s new local-first AI runs on your PC — and asks before using the cloud
• ChatGPT can learn your repetitive tasks by watching you — here are 7 prompts to try
• Google just gave Gemini access to your ebooks — 7 prompts for readers and students

I’ve found so many ways to push myself and put my lazy ways behind me by implementing productivity routines into my daily schedule.

And with so many AI tools at my disposal, I’ve discovered a wealth of productivity-boosting hacks that have made me feel and work better. Completely disconnecting from physical & digital distractions, moving around as much as possible & staying hydrated during my work break and writing up digital journal entries with notes on everything I accomplished & hope to complete next have worked out great for me thus far.

Since I’m always on the lookout for productivity methods that can keep me in a focused and efficient state, I went to ChatGPT, Gemini, Claude and Perplexity with a new mission: discover some of the more unconventional productivity boosters that I can try out for myself.

All four chatbots spit out a bunch of information about worthwhile techniques that worked as well as advertised.

Productivity practices I learned about for the very first time