If you use AI chatbots for anything remotely sensitive — finances, health questions, job concerns, parenting, relationships — choosing the right one matters more than you think. Every LLM handles your data differently. Some save your chats forever. Some send them to human reviewers. One doesn’t train on your data at all.
I dug through the privacy practices of the four biggest AI chatbots — OpenAI’s ChatGPT, Google’s Gemini, Anthropic’s Claude and Perplexity — to see what really happens to your conversations behind the scenes. The differences were bigger than I expected.
Here’s how each chatbot treats your data and which one is the safest choice if privacy is your top concern.
ChatGPT: the least private AI
ChatGPT is the most widely used chatbot — and also the least private unless you actively change your settings. By default, ChatGPT uses your conversations to train its models. Your chat history is saved indefinitely unless you manually delete it.
OpenAI does offer Temporary Chat, which prevents your conversation from being saved or used for training. But you have to remember to turn it on each time you need it.
To OpenAI’s credit, the controls are there, but privacy requires deliberate effort. For everyday use that isn’t sensitive, it’s great. For private matters, you may want to think twice.
Why ChatGPT ranks last
- Uses your chats for training unless you opt out
- History is stored indefinitely
- Deleting history requires manual cleanup
- Private modes aren’t the default
Gemini: Good controls, but human reviewers can see your chats
Gemini is more complicated because it sits inside your Google account, which gives you strong privacy controls, but also means the default behavior leans toward retention and training.
The biggest privacy concern is this: some Gemini chats are routinely sent to human reviewers, and these reviewed interactions can be retained for up to three years, even if you disable data saving or delete your history.
You can manage these settings under Google Account > Data and Privacy > “Gemini Apps Activity,” where you can disable activity logging or shorten auto-delete windows.
Gemini is fine for casual use, but I wouldn’t put sensitive personal details into it without adjusting settings first.
Why Gemini ranks No. 3
- Full ability to disable activity logging
- Auto-delete can be set to 3 months
- Deeply integrated privacy controls
- But: Human reviewers may read selected chats
- Reviewed chats can be stored for up to 3 years
- Defaults favor retention and training
Perplexity: strong controls and an easy opt-out
Perplexity comes second because it gives users the most granular controls. By default, Perplexity may use your data for training — but the toggle to disable this is simple and easy to find under settings (“AI Data Usage”). If you want a completely private session, Incognito Mode ensures nothing is saved at all.
You can also clear or disable memory, giving you full control over what the system remembers about you.
Its only drawback is that training is still enabled by default. But unlike many platforms, Perplexity actually makes the opt-out obvious.
Why Perplexity ranks No. 2
- Clear and simple opt-out from training
- Incognito Mode ensures nothing is stored
- Memory can be fully disabled or cleared
- Transparent data practices
Claude: the most private chatbot by far
Claude is the only mainstream chatbot that doesn’t train on your conversations unless you explicitly opt in. That means your chats aren’t quietly used to improve the model, and they aren’t fed into large training pipelines unless you choose otherwise. If you delete a conversation, Anthropic removes it from their systems within roughly 30 days.
There’s also no routine human-review pipeline for your chats. Unless you manually submit feedback, your data stays private. Claude is hands down the best choice for anyone who wants maximum privacy without tinkering with settings.
Why Claude ranks No. 1 for privacy
- Doesn’t train on your conversations by default
- Minimal data retention
- Deleted chats removed within ~30 days
- No hidden human review unless you volunteer data
Verdict: Which chatbot should you trust?
If your priority is privacy above everything else, the ranking is clear:
- Claude — best for privacy
- Perplexity — flexible and user-controlled
- Gemini — fine with the right settings, but human review is a concern
- ChatGPT — powerful but data-hungry by default
For the most private choice with the fewest compromises, Claude is the clear winner. If you want tight control and customizable privacy tools, Perplexity is excellent. If you're deep in the Google ecosystem and willing to adjust settings, Gemini works but has strings attached. If you're using ChatGPT regularly for sensitive issues, switch to Temporary Chat or disable training before you share anything personal.
Which chatbot do you prefer? Is privacy one of the features you take into consideration? Let me know in the comments.
More from Tom's Guide
- 11 underrated AI tools that can save you serious time — and most are free
- I found 5 shockingly helpful uses for ChatGPT you haven't tried yet — and they all make my life easier
- I tested 7 money-saving prompts on ChatGPT-5.1 and Gemini 3.0 — here's the surprising winner
Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds.
You must confirm your public display name before commenting
Please logout and then login again, you will then be prompted to enter your display name.
