ChatGPT has been taking the world by storm since launching in late 2022, and it is easy to see why. The revolutionary chatbot AI can do a surprising amount of tasks, from holding a conversation to writing an entire term paper. Plus, there are a lot of things you didn't know that ChatGPT can do — from making a brand logo to composing music and more.
We know that lots of people are trying to figure out how to use this ChatGPT and what its limitations are. If you want to know how to use this chatbot AI check out our guide on how to use ChatGPT, as well as these tips to get the most out of ChatGPT. Here we answer all your top questions about ChatGPT.
ChatGPT news and updates (Updated September 27)
- Jony Ive could be teaming up with ChatGPT — what we know
- Meta AI chatbots unveiled — Tom Brady, Kylie Jenner, Snoop Dog and more
- Windows 11 update with Copilot just dropped — here's what it's like to use Microsoft's AI assistant
What does ChatGPT stand for?
According to OpenAI, ChatGPT is, "an artificial intelligence trained to assist with a variety of tasks." More specifically, though, it is a language model AI designed to produce human-like text and converse with people, hence the "Chat" in ChatGPT.
Practically, this means that to use ChatGPT, you present the model with a query or request by entering it into a text box. The AI then processes this request and responds based on the information that it has available.
Generate written content from news articles to novels
Summarize long documents
Answer questions as a research tool
Write and debug code
Build and text-based games
Act as a tutor for homework questions or problems
Plan your next vacation
Create software activation keys
The "GPT" in ChatGPT comes from GPT, the learning model that the ChatGPT application utilizes. GPT stands for Generative Pre-trained Transformer and most people are currently using GPT-3.5. This is the version of GPT that is powering the free research preview version of ChatGPT.
There is a newer model as well, called GPT-4. However, this model is only available to ChatGPT Plus subscribers and developers using the GPT-4 API. This may eventually change, but for now, free users are stuck with GPT-3.5.
In an August 2023 update, OpenAI added additional functionality to ChatGPT. The AI chatbot will now offer prompt examples when you start a chat and suggest replies within conversations. It now also defaults to GPT-4 if you have ChatGPT Plus, though GPT-3.5 is still available. And Plus users also are able to upload multiple files at once for ChatGPT to analyze. The update also enabled keyboard shortcuts and no longer automatically logs you out.
ChatGPT isn't reserved for just text on a screen though. In a September 2023 update, OpenAI added the ability for ChatGPT to respond to you by talking to you. And you can ask it questions by talking to it as well. The feature is currently limited to ChatGPT Plus and Enterprise users, but may eventually make its way to the free tier.
Can you use ChatGPT for free?
ChatGPT is still available to users as a free service in the research stage. Just create an account — which you can learn how to do in our guide to ChatGPT — and you're good to go.
However, OpenAI has also launched ChatGPT Plus, a paid subscription service for ChatGPT. It costs $20 a month and promises access to ChatGPT even when demand is high, faster response speeds and priority access to new features when they become available. It also includes access to the new GPT-4 large language model.
Despite these promises, some users have still complained about being unable to use ChatGPT due to the service being at capacity. So make sure to check out our guide to everything you need to know about ChatGPT Plus before subscribing.
However, these users are reaping some benefits from their membership. OpenAI has announced a beta feature for ChatGPT Plus users that gives them access to ChatGPT Plugins and a new version of ChatGPT that can browse the internet. We tried a ton of ChatGPT plugins and found a few that stand above the rest. They even make the $ 20-a-month fee worth considering.
Plus ChatGPT Plus users also got early access to Custom instructions, which tailors ChatGPT to your needs. This feature allows you to tell ChatGPT who you are and how you want it to respond, making using the chatbot a more efficient experience. OpenAI has now made Custom instructions available to all U.S. users, with additional countries expected to get access in the future.
There's also the ChatGPT API, which developers use to integrate ChatGPT into apps. However, you can also utilize this API to do a wide range of things, including putting ChatGPT on your iPhone through a shortcut. The API is typically a pay-as-you-go product, with pricing starting as low as $0.002 per 1,000 tokens (a token is pieces of words that the chatbot uses to process prompts).
Oh, and you can even get paid to use ChatGPT — though it's a full-time job. Job listings for Prompt Engineers are popping up and some pay over $300,000 a year. So if you're passionate about AI, that may be a career you wish to explore. You can also get paid to find and report ChatGPT bugs through an official bug bounty program if that's more your speed.
Why did ChatGPT get banned?
ChatGPT was banned in Italy for a while. The Italian government issued a government order that banned ChatGPT on the grounds that OpenAI unlawfully collected personal data. There were also concerns over OpenAI not requiring an age verification system.
But that ban has since been lifted. As reported by The Wall Street Journal, OpenAI made sufficient concessions in the eyes of Italian regulators, so the chatbot is back online in the country.
Currently, the only countries where ChatGPT is unavailable are China, Russia, North Korea and Iran, where OpenAI has declined to make the service available.
However, more countries could follow suit if Eric Schmidt's latest comments are any indication. The ex-Google CEO warned of an AI 'reckoning' in a recent interview, stating that it could ultimately undermine democracy. And "The Godfather of AI" Dr. Geoffrey Hinton recently resigned from Google because they could no longer support working on AI, furthering the case being made by some for pressing pause on AI.
Given what we've seen with the potential TikTok ban, chatbots like ChatGPT are likely going to be perceived as a threat by governments and that means they will probably take it seriously.
And corporations are taking the threats of using AI seriously too. Samsung just banned employees from using ChatGPT and other generative AI after employees accidentally leaked sensitive information to the AI chatbot. However, while Samsung has banned the internal use of ChatGPT, it's still looking at generative AI. Bixby voice assistant and the S-Pen could be prime candidates for an AI upgrade based on comments from Samsung executives.
Apple has also followed suit, banning its employees from using ChatGPT despite having just released an iOS app for the popular chatbot. Apple even internally created its own chatbot "Apple GPT" so that it wouldn't need to use ChatGPT for its generative AI needs.
Valve, the company behind the popular PC gaming platform Steam, is also giving AI the ban hammer. Valve is blocking AI-generated content on Steam, but not because it's taking a stance against AI. Rather, the company is seemingly trying to steer clear of any legal issues and is “working through how to integrate it into [its] already-existing review policies.” So AI-generated content may eventually make its way onto Steam provided it can be proven not to violate copywrite or cause similar issues.
Can people detect if you use ChatGPT?
As ChatGPT becomes more prevalent in writing, people are starting to create AI tools to detect ChatGPT or similar AI models in written content.
GPTZero is one such tool, created by Princeton University student Edward Tian. According to NPR, GPTZero uses “perplexity” and “burstiness” scores to measure the complexity of text. GPTZero was able to differentiate between an article from The New Yorker and a LinkedIn post written by ChatGPT, so there’s some early evidence that it works at detecting the use of ChatGPT.
The theory behind these tools is that humans write in a way that is more complex than content written by other AI. We even tested whether ChatGPT will steal our jobs and all four of our staff testers were able to tell what reviews were written by humans and which were written by ChatGPT. You can try and teach ChatGPT your writing style, but even then it could still be detected.
Don't believe us? Black Mirror series creator Charlie Brooker asked ChatGPT to write a Black Mirror episode and it was "s**t." Specifically, Brooker said that while the chatbot can synthesize previous Black Mirror episodes and produce a plausible script, a deeper look shows that the AI-created episode lacks any original thought.
And Black Mirror isn't the only show to get the AI treatment. A company called Fable Studios created an entire fake South Park episode using AI, though it says it won't release the ability to create South Park episodes as it's not its IP. Instead, it's releasing SHOW-1, an AI tool that can write scripts, animate scenes, direct, do voiceover work and edit for you.
Additionally, ChatGPT can plagiarize without you knowing. Since ChatGPT pulls data from all over the internet and beyond as part of its model training, it pulls in data is not considered common knowledge. If you include something in a written work and it is not considered common knowledge or you are not the primary source, you need to cite it to avoid plagiarism. While the chatbot can provide quotes, and in some cases even fool plagiarism checkers, you need to be vigilant when using the chatbot to avoid plagiarism.
OpenAI is currently in hot water because of this issue. They are now facing a class-action lawsuit over how they use data in ChatGPT. The basis of the legal argument is ChatGPT summarized the works of two authors so precisely that the chatbot AI must have been trained on the works, allegedly without permission. If that was the case, it may turn out that a lot of ChatGPT's training was done without permission from authors, but we won't know for certain until the legal battle plays out.
Is there a ChatGPT app?
There is an official ChatGPT app! While for a long time, you had to settle for using a web browser, OpenAI has now launched an app that lets you do everything that ChatGPT can do on a web browser. And there's a ChatGPT Android app now too. While limited to the U.S. initially, access to the official iOS app has expanded since its launch. The Android app currently requires you to pre-register for access.
Aside from the official app, there are also other apps using ChatGPT. Snapchat now has My AI, which is ChatGPT integrated into the popular messaging app. Opera has also integrated ChatGPT into its web browsers, allowing users to summarize articles and web pages, generate social media posts and more, with just a prompt or click. Even Slack has integrated ChatGPT into Slack's app. But none of these is a standalone ChatGPT app.
And be careful of apps claiming to be ChatGPT apps. Fake ChatGPT apps are spreading malware that can steal your money and passwords. Others are scamming users out of thousands of dollars. So if you want to use ChatGPT on your phone, you can either do it through your mobile browser or use the official iOS app.
ChatGPT-3 vs ChatGPT-4
With the launch of GPT-4, a lot of people have been wondering what differences there are between ChatGPT and GPT-4. The terminology can even be confusing, with terms like ChatGPT-3 and ChatGPT-4 — and now ChatGPT-5 — being thrown around. So first, let's cover the different terminology.
OpenAI has been around since 2015 and has been working on the GPT model behind ChatGPT for most of that time. Then in late 2022, they launched ChatGPT, the popular chatbot that we've now become so familiar with. This ChatGPT chatbot was powered by GPT-3.5, an updated version of the GPT-3 model that was the third iteration of the GPT large language model.
So if anyone mentions ChatGPT, ChatGPT-3, ChatGPT-3.5 or GPT-3.5, they are talking about the free version of ChatGPT and/or the language model powering it.
Then there is GPT-4, which is the latest version of the GPT model. Sometimes referred to as ChatGPT-4, this model launched in March 2023 and is an upgraded version of ChatGPT. Currently, only those with ChatGPT Plus or developers with access to the ChatGPT API (more on that later) have access to this new model. However, it has powered the ChatGPT integrations in other apps such as the new Bing with ChatGPT, so while most people don't have direct access to ChatGPT-4, you could be using it in other apps.
But what does this mean practically? In short, GPT-4 is a massive leap forward. It processes things faster, can process more lines of text and can even process images and provide context on those images. However, there are hacks that can be used to get around ChatGPT-3.5's limits, such as using "Shogtongue," a language that ChatGPT created to allow conversations to go on longer than the 8,000-word limit.
GPT-4's ability to handle both text and images is called multimodal functionality, and if you ever read someone talking about that, they are talking about ChatGPT-4. ChatGPT-3 and ChatGPT-3.5 are text-based only — even though you can use the text code provided by ChatGPT to create images and 3D models.
So to sum this all up, ChatGPT (aka ChatGPT-3) can take in text input and create text outputs. It's powered by GPT-3.5. ChatGPT-4 (aka GPT-4) is an upgraded version that is much more powerful and can also handle images as inputs but is limited to ChatGPT Plus users and developers.
What is ChatGPT 5?
ChatGPT-5 — or GPT-5 — is the rumored next version of ChatGPT's GPT model. It was rumored to be ready around December 2023, but that rumor has since been debunked. At an MIT event, OpenAI founder Sam Altman said that OpenAI is not working on GPT-5 and "won't be for some time." Regardless of when it's coming, we still don't know much about it yet.
The one thing we have heard rumored though is that ChatGPT-5 could achieve artificial general intelligence (AGI). This means it could pass the Turing test, which is a test that determines if a computer can communicate in a manner that is indistinguishable from a human. Google DeepMind CEO Demis Hassabis has said he expects we’re ‘just a few years away’ from human-like AI.
This could be a revolutionary step forward or a step too far depending on who you ask. Several tech leaders have called on ChatGPT and Google Bard to halt AI training out of concerns for safety. OpenAI seems to show no desire to stop training ChatGPT, however, so time will tell whether ChatGPT-5 brings about a scenario where the machines rise up and take over or merely becomes an incredibly powerful AI tool.
What is the ChatGPT API?
Throughout this article, you've heard mention of the ChatGPT API. An API, or application programming interface, is a tool that developers can use to integrate ChatGPT into their own apps.
This has already led to a variety of applications, including the Amazfit GTR4 smartwatch, which claims to take the capabilities of ChatGPT's AI model and put it in a smartwatch. Snapchat's My AI and the Slack ChatGPT app are both prime examples of the ChatGPT API in use, but there are many more.
So if you're using an application that has ChatGPT features built in, that is likely the ChatGPT API. The one thing to note here is that the ChatGPT API uses GPT-4 rather than GPT-3.5, so apps using the ChatGPT API could be more powerful and have greater functionality than the free version of ChatGPT.
One final note: the ChatGPT API is different from ChatGPT plugins. The API brings ChatGPT's tools to other sites, whereas the ChatGPT plugins take other sites and add their functionality into ChatGPT.
One example is Expedia's ChatGPT plugin, which allows you to ask ChatGPT to plan a vacation and ChatGPT would pull from Expedia to help you do things like book flights and hotels.
OpenAI has admitted that there may be safety concerns with these plugins but they are implementing precautions and transactional information. For example, purchases will be kept separate from the plugin. Keep an eye out for new plugins coming to more and more of your favorite sites.
If you want to access ChatGPT plugins check out how to use ChatGPT web plugins. The plugins aren't widely available to everyone just yet so join the waiting list if you're interested. If you want to see how a generative AI could use plugins in the meantime, Microsoft Bing introduced plugins with sites such as OpenTable in a slew of new Bing Chat upgrades. All you need to test those plugins is a Microsoft account.
There are quite a few ChatGPT alternatives, but the biggest competitor has to be Google Bard. Bard is similar to ChatGPT in that it is a chatbot that can answer complex questions, generate content like poems and emails and help you plan a party or vacation. And with a recent update, Bard can even write and debug code in over 20 programming languages from C++ to Python.
However, Bard is a standalone tool right now that is separate from Google Search, although it may be integrated in the future. It's already been integrated into other Google apps like Google Sheets, which is a serious time saver.
Google is also reportedly heavily invested in Anthropic, a rival to OpenAI. Google is said to have invested $400 million in Anthropic. Anthropic unveiled its GPT-4 competitor — Claude 2 — in July 2023.
And now there is Google's Search Generative Experience — or SGE. It's a version of Google Search that integrates a generative AI chatbot into the top of the Google Search results page, replacing the traditional list of blue links with an AI-generated response. Check out our guide on how to sign up for Google Search Labs if you want to sign up for SGE access and try the future of AI search engines for yourself.
But the next big thing could be a Google AI project called "Gemini." This AI is being developed by Google DeepMind, and DeepMind CEO expects it to be a multimodal AI model that competes directly with GPT-4. “Gemini was created from the ground up to be multimodal, highly efficient at tool and API integrations and built to enable future innovations, like memory and planning. While still early, we’re already seeing impressive multimodal capabilities not seen in prior models.”
Aside from what Google is working on, there is also You.com's AI chatbot, which is a multimodal search engine with chatbot functionality. It's not bad as a search engine, but it lacks the capabilities of ChatGPT when it comes to content creation or coding. For those in China, Alibaba has also unveiled its ChatGPT competitor called Tongyi Qianwen and Baidu has its own chatbot called Ernie.
On the more novel side, Stability AI — the company behind the AI image generator Stable Diffusion — has launched its own open source ChatGPT competitor called StableLM. It's not as well-trained but the potential is something to keep an eye on. For something on the whimsical side, there's CatGPT, which is ChatGPT but with the chatbot responding as a cat.
And if none of that appeals to you, you could always just wait and see what Elon does. The Twitter and Tesla CEO is rumored to be working on his own AI chatbot through the company X.AI that he founded back on March 9, 2023.
But the most interesting ChatGPT alternative might be Auto-GPT, which is a version of ChatGPT that uses a Python environment to automate a lot of the follow-up prompts required to get the best response from ChatGPT. It takes a bit of expertise to work, but the efficiency it provides is well worth it.
Keep an eye on Meta too. The company plans to bring AI to billions in Facebook, Instagram and WhatsApp based on a recent earnings call, with generative AI almost certainly making its way into some of the company's apps. They also released Llama 2 in partnership with Microsoft, making a powerful LLM available for free. Llama 2 was used to create Meta's AI chatbots, which are all portrayed by celebrities and help you with specific tasks rather than anything and everything.
And not to be left out, Amazon has finally unveiled its first major foray into consumer-facing generative AI. At its September 2023 event, Amazon announced a new version of Alexa powered by generative AI. This newer version should be more conversational and can be personalized to remember things about you such as favorite sports teams and more.
Even Tubi is getting into the game. Tubi has added an AI chatbot dubbed Rabbit AI and it will help you find a movie when you have an idea of what you want to watch but don't have a specific title in mind.
If you want to skip over text-based generative AI, we've got you covered too. Our guide to the best AI video generators has five excellent picks that we've tested so you know what the best option is for you if you want to create videos using AI technology. One offering that might make the cut soon? Runway. Its Gen 1 AI could use AI to edit existing video clips, but Runway Gen 2 allows you to create new AI videos with just a text prompt. We even faced off Runway Gen 2 against the best AI synthetic video generators — these are AI video generators that create video clips without needing existing video content. Prefer images to videos? Check out Midjourney, a popular AI image generator.
And using text prompts to create new multimedia content isn't limited to video and images either. Meta has a text-to-audio AI tool called MusicGen, an open-source tool that converts text prompts and melodies into new 12-second music clips. Los Angeles AI company HeyGen has also created its own AI audio tool called Video Translate. This tool takes a video of you speaking in in English, Spanish, French, Chinese, German, Italian, Portuguese, Dutch, Hindi or Japanese and translates it into one of those same seven languages.
The lone company that may stay away from the AI gold rush is Apple. There are rumors that Apple could have a generative AI-based overhaul of Siri as early as 2024, but don't expect ChatGPT-like features from Siri anytime soon. Instead, expect Apple to use machine learning behind the scenes to improve its products.
- Read more: 5 astonishing uses of AI happening right now
What is Bing with ChatGPT?
Aside from the ChatGPT alternatives above, there is also "the new Bing" — or Bing with ChatGPT.
Bing with ChatGPT was announced at a Microsoft event on February 7, 2023. and it is closer to a GPT-powered search engine than a ChatGPT competitor. It is powered by the same GPT model that powers ChatGPT, though it uses GPT-4 rather than GPT-3.5. And Microsoft seems to be all in on Bing Chat and its other AI projects, as it has officially killed off its old AI assistant Cortana.
We were initially impressed with its potential but as we've spent extended time with the Microsoft chatbot issues crop up. Sometimes we even wonder how can ChatGPT be the next big thing if it's already breaking when pushed beyond basic requests. It often gets the basic stuff wrong at a surprising rate — though this happens with chatbot AI more than you'd think.
Despite this, the new Bing seems to be a winning strategy for Microsoft. The company says that since the new Bing was unveiled in February, it took less than a month for Bing usage to swell to over 100 million daily active users. It will be interesting to see if that active user number increases as ChatGPT search results become more prevalent in Bing. This success already reportedly to has Samsung questioning if it needs to ditch Google for Bing on its Galaxy phones.
The GPT-powered Bing was initially available only through a waitlist but has since been made open to everyone. Well, everyone with a Microsoft account. And you're no longer limited to Microsoft's Edge browser! Bing Chat is now available on Google Chrome and some users are even getting access on Safari.
Aside from web browsers, Bing is now on a lot of apps — and I mean a lot. As of February 22, Microsoft's new Bing chatbot is available on the iOS and Android Bing, Edge and Skype apps. Microsoft also added Bing with ChatGPT on its SwiftKey keyboard on Android and SwiftKey for iPhone and has created a new Bing Chat widget for Android and iOS. It even added it to the Windows 11 taskbar — check out our guide to enable (or disable) ChatGPT on the Windows 11 taskbar if you want to remove Bing Chat from Windows.
In addition to making Bing available in as many places as possible, Microsoft has also supercharged Bing with ChatGPT with upgrades. Microsoft has introduced plugins into Bing Chat, allowing you to make a dinner reservation straight from the chatbot. You can also have the chatbot open a tv show or movie in a streaming service just by asking it, which could be highly useful for mobile users. Aside from these changes, there are also user interface upgrades, the addition of chat histories and more. These features are rolling out now, so if you don't have access to them yet don't panic.
Make sure to check out our guide to nine practical uses of Bing with ChatGPT. It's a great tool for using the chatbot for things it's actually good at. We also have a guide on how to save or export your Bing Chat responses, and on the lighter side, check out our guide to games you can play using Bing right now.
And if you want generative AI upgrades infused into your Windows 11 experience, Windows 11 Copilot takes generative AI and puts it in the popular operating system. The AI assistant is available as of September 26 and while we weren't initially convinced by its potential, seeing it up close and personal has us convinced Copilot isn't another AI gimmick.
- For more on how Microsoft's Bing with ChatGPT compares against another AI — Google Bard — check out our Bing with ChatGPT vs Google Bard face-off.
What is Apple GPT?
Apple GPT is a generative AI chatbot, much like ChatGPT. However, unlike ChatGPT, it does not use the GPT-4 LLM. Instead, it was built on a framework called "Ajax," which is based on Google's machine learning framework Jax, and it runs on Google Cloud servers. The GPT in Apple GPT is simply a nod towards ChatGPT's current market dominance and the similarities between the two generative AI chatbots. They are not affiliated with each other in any way.
In terms of what it can do, the features are very similar to the basic features of ChatGPT. Apple GPT can summarize texts and answer questions based on available data. Apple employees are largely using it to assist in prototype testing.
Here's the thing — you may never use Apple GPT. At the moment, it's a tool for Apple employees only, and they need special permission to use it. They cannot even use outputs from Apple GPT in consumer-facing products. But it's been rumored that Apple will make a big AI announcement sometime in 2024, so Apple GPT could ultimately be involved in that larger announcement.
Why is ChatGPT at capacity?
ChatGPT has constraints in terms of how much it can process at once, so it throttles the number of users that can access it at any given time.
This is the most common reason that it will not work — if ChatGPT is at capacity, it will not let you log in. One of the big selling points of ChatGPT Plus mentioned earlier is priority access to ensure you don't encounter this issue, though ChatGPT Plus users have still reported getting the error message that ChatGPT is at capacity.
However, ChatGPT could get increased bandwidth if Microsoft is successful in building its own AI chip. Microsoft has been reportedly working on its own AI chip for some time, and it is hoping that it can be mass-produced as soon as 2024. If successful, these chips could power the Azure supercomputers that power ChatGPT, and could greatly alleviate bandwidth bottlenecks.
Aside from this roadblock, ChatGPT can still suffer from technical errors like any other site or app. It can have server errors preventing it from working, or if you have a poor internet connection you may struggle to use it successfully.
Is ChatGPT safe?
This is a complicated question. In one sense, yes, ChatGPT is safe. If you log into your OpenAI account and use it, it won’t install anything malicious onto your device.
However, you still need to be concerned about OpenAI suffering a data breach and exposing your personal data, which is a risk with any online account. We've already seen that happen to a small number of ChatGPT Plus users who were affected by a bug that exposed "user’s first and last name, email address, payment address, the last four digits (only) of a credit card number, and credit card expiration date." While only a small percentage of users were affected, this shows that OpenAI still suffers from the same security risks as any website.
There has also been a breach where over 100,000 ChatGPT accounts were compromised due to malware. These account credentials were subsequently put on the dark web for sale to malicious actors. The ChatGPT site itself wasn't the cause of this breach, but you should still always be careful when using it.
We've also learned that ChatGPT may be sharing your secrets with how it records your chat history. Check out our guide on how to disable Chat History and Training on ChatGPT so you can limit what ChatGPT remembers.
Speaking of bugs, OpenAI has introduced its own Bug Bounty Program, challenging users and ethical hackers to report any issues they find, with some potentially big money rewards, up to $20,000. This will hopefully see ChatGPT become more secure than ever before.
On top of account safety concerns, you need to be conscious of what data you put into ChatGPT regardless of your account type. According to OpenAI’s ChatGPT FAQs article, ChatGPT does save your conversations and they are reviewed by OpenAI for training purposes. Recently a bug caused those conversation histories to be visible to other users, forcing ChatGPT to disable the feature for a short time. Samsung also found out that ChatGPT stores your data the hard way, as it accidentally leaked its secrets to ChatGPT multiple times by using ChatGPT to optimize tests for its chips, among other things.
If you want to delete your data, you’ll have to delete your entire account, which is irreversible. To do so, just go to this OpenAI help page and follow the instructions.
Additionally, with AI there are deeper ethical and moral concerns — especially since the AI model has neither ethics nor morals. As Bleeping Computer lays out, ChatGPT can be unknowingly offensive in its responses, breed misinformation, write phishing emails, be sexist, racist, etc. Because the AI model pulls information from the internet and other sources to form its knowledge base, it can potentially pull the harmful stuff without knowing that it's harmful. So just be mindful of this lack of safeguards when using the service.
Finally, there are some mental health safety concerns with using AI that can sometimes go off the rails. Some AI experts have proposed digital health warnings for chatbot AI like ChatGPT and even Apple appears to be banning apps using the ChatGPT API over safety concerns.
Oh, and don't forget DAN (AKA Do Anything Now). DAN is the alias for the jailbroken version of the chatbot, some are describing it as ChatGPT's evil twin. Not approved by OpenAI, DAN is essentially ChatGPT being tricked into assuming a persona that bypasses its terms of service in order to respond to prompts asking it unethical, violent or offensive questions. OpenAI is working constantly to stop DAN from being accessed and if you use it, you do so at your own risk. Even without using DAN, ChatGPT can be used to create malware. In fact, if you have a Mac, there's a new ChatGPT-created dark web hacking tool that can give bad actors complete access to your Mac. The good news is that this tool is incredibly expensive, so your personal Mac is unlikely to be targeted. But it still shows how dangerous ChatGPT can be when the guardrails are taken off.
There's also now another unsafe ChatGPT clone — WormGPT. This AI chatbot has "no ethical boundaries" and is designed specifically for criminal purposes.
And while ChatGPT wasn't used in this instance, a recent fake kidnapping where AI faked a daughter's voice to extort her mother highlighted the dangers of AI beyond these chatbots. So while ChatGPT is by no means malicious content, calling AI as a whole "safe" may be a bridge too far.
How to keep your ChatGPT chats private
By default, all of your conversations with ChatGPT will go to OpenAI to help train the AI and improve the model. If you'd rather have more privacy, you can now enable the equivalent of a search engine's Incognito mode to stop ChatGPT sharing your secrets. This is a simple process and while your chats will still be stored on ChatGPT for 30 days, they will then be deleted instead of being used to train the AI.