The key AI privacy settings are one switch per app. ChatGPT: Settings → Data controls → Improve the model for everyone, off. Gemini: turn off Keep Activity in Gemini Apps Activity. Claude: Settings → Privacy, then turn off model improvement. Copilot: turn off Training on conversation activity and Training on voice conversations under Privacy.
Meta AI has no training opt-out for US users, so delete your Meta AI chats and don’t share anything personal. None of these switches delete chats you already have.
We checked every path below against each company’s own help pages on October 3, 2026. Menus change often, so if a label looks slightly different on your device, look for the closest match.
AI privacy settings at a glance
| Assistant | Switch to turn off | Where it is | Private chat option | Kept after opting out |
|---|---|---|---|---|
| ChatGPT | Improve the model for everyone | Settings → Data controls | Temporary Chat | Temporary chats up to 30 days for safety; deleted chats removed within 30 days |
| Gemini | Keep Activity | Gemini Apps Activity | Temporary chat | Chats kept up to 72 hours; human-reviewed chats up to 3 years |
| Claude | Model improvement setting | Settings → Privacy | Incognito chat | Deleted chats removed from back-end storage within 30 days |
| Microsoft Copilot | Training on conversation activity and Training on voice conversations | Profile → Privacy (exact path varies by app) | — | Conversation activity stored 18 months by default unless you delete it |
| Meta AI | No training opt-out in the US | Meta AI app → Settings → Data & privacy | None | Until you delete chats |
How to stop ChatGPT using your data
- Web: open your account menu → Settings → Data controls, select Improve the model for everyone, turn it off and select Done.
- Android and iPhone: open the sidebar, tap your profile icon to open Settings → Data controls, and turn off Improve the model for everyone.
- When you’re signed in, the setting follows your account to every device. Signed-out users can change it too, but it’s saved only in that browser.
For a one-off private conversation, start a Temporary Chat. It doesn’t appear in your history, doesn’t create memories and isn’t used for training, though OpenAI may keep it for up to 30 days for safety checks. Two more details matter:
- Thumbs up or down counts as feedback. OpenAI says that if you rate a response, the whole conversation may be used for training even after you opt out.
- Ads are a separate setting. On the Free and Go plans in the US, ads can appear below answers. Turn off ad personalization in Settings → Ad controls. OpenAI says it doesn’t share your conversations with advertisers or sell your data to them. See ChatGPT Free vs Plus for which plans have ads.
Gemini: turn off Keep Activity
- Android app: open Gemini, tap Menu → your profile picture → Gemini Apps Activity.
- Computer: go to gemini.google.com → Settings & help → Activity, or open myactivity.google.com/product/gemini.
- Near the top, tap On, then Turn off or Turn off and delete activity.
- Optional: change Auto-delete from the default 18 months to 3 months if you leave Keep Activity on.
If you’re 18 or over, Keep Activity is on by default. With it on, Google uses your chats, uploads and related data to improve its services, including training its AI models, and a subset is read by human reviewers. Reviewed chats are disconnected from your account and kept for up to three years, even if you delete your activity.
With Keep Activity off, Google still keeps chats with your account for up to 72 hours to answer you and for safety. Turning it off also limits which connected apps Gemini can use. Recordings from your voice, Gemini Live video and screen sharing are only used for training if you opt in separately. More on Gemini’s settings in how to use Gemini on Android.
Claude: turn off model improvement
- On the web or desktop app, select your name → Settings → Privacy.
- On the mobile app, select your name → Settings → Privacy.
- Turn off the switch for helping improve Anthropic’s AI models (Anthropic’s help page calls it Help Improve our AI models).
Once it’s off, Anthropic says it won’t use your new or previously stored chats in future training runs, though anything already used in training can’t be removed. Incognito chats are never used for training. Two exceptions: chats flagged by safety systems may still be used to improve safety models, and if you rate a response, that conversation is kept for up to five years.
Microsoft Copilot: turn off training on text and voice
- copilot.com: select your profile icon → your profile name → Privacy.
- Windows or Mac app: select your profile icon → Settings → Privacy.
- Mobile app: open the menu, select your profile icon → Account → Privacy.
- Turn off Training on conversation activity and Training on voice conversations.
Microsoft released a new version of the Copilot app on August 18, 2026, and says its older privacy instructions apply to the previous version.
Microsoft is clear about the limits: opting out stops future conversations being used to train its AI models, but not their use for general product improvements, advertising, safety and security. Personalization and memory are separate settings, and personalized ads are controlled in your Microsoft account. Microsoft says it doesn’t train on conversations from signed-out users, users under 18, or work accounts.
Meta AI (Facebook, Instagram, WhatsApp, Messenger)
Meta offers the fewest controls. Since December 16, 2025, Meta has used your conversations with Meta AI, typed or spoken, to personalize the content and ads you see in most regions, and there’s no switch to turn that off. Meta says it doesn’t use sensitive topics such as health, religion or political views for ads.
- To delete all your Meta AI chats in the Meta AI app, tap Menu → Settings → Data & privacy → Manage your information → Delete all chats and media.
- Review Ad preferences in Accounts Center to limit how ads are targeted.
- In Facebook, Instagram and WhatsApp, avoid tagging @Meta AI in chats you want to keep private.
The safest rule for Meta AI is simple: don’t tell it anything you wouldn’t post publicly.
What these settings don’t do
- They don’t delete old chats. Delete chats separately, or delete your whole history.
- They don’t stop safety retention. Every company keeps some chats for a period to detect abuse, even when training is off.
- They don’t undo past training. Data already used in a trained model stays there.
- They don’t cover work or school accounts. Your organization controls those settings; OpenAI and Microsoft say they don’t train on business accounts by default.
- They’re not the same as memory. Memory and personalization are separate settings in ChatGPT, Gemini and Copilot.
Which AI assistant is most private?
Going by the settings alone, ChatGPT and Claude are the simplest: one switch stops training while keeping your chat history, and both have a private chat mode. Gemini’s Keep Activity switch also turns off your saved history, which some people will find a trade-off. Copilot’s opt-out leaves room for advertising and product-improvement use, and Meta AI has no training opt-out for US users at all.
Comparing assistants on more than privacy? See ChatGPT vs Gemini vs Claude vs Copilot. On a Samsung phone, Galaxy AI has separate switches: how to turn off Galaxy AI. More guides in our AI hub and security hub.
FAQ
How do I stop ChatGPT from using my data for training?
Go to Settings → Data controls and turn off Improve the model for everyone. New chats won’t be used for training, but avoid the thumbs up/down buttons, because rated conversations may still be used.
Does turning off training delete my chats?
No. It only affects future use for training. Delete old chats separately in each app.
Is Temporary Chat in ChatGPT private?
It isn’t saved to your history, doesn’t update memory and isn’t used for training, but OpenAI may keep it for up to 30 days for safety checks.
Can I stop Meta AI using my data?
Not fully in the US. There’s no training or ad-personalization opt-out for Meta AI chats, but you can delete your chats and limit ad targeting in Ad preferences.
Does ChatGPT sell my data to advertisers?
OpenAI says it doesn’t share your conversations with advertisers and never sells your data to them. On Free and Go, you can turn off ad personalization in Settings → Ad controls.
Sources
- Data controls in ChatGPT — OpenAI Help Center
- Ads in ChatGPT — OpenAI Help Center
- Gemini Apps Privacy Hub — Google
- Manage and delete your activity in Gemini Apps — Google
- How do I change my model improvement privacy settings? — Anthropic Privacy Center
- How long do you store my data? — Anthropic Privacy Center
- Microsoft Copilot privacy controls — Microsoft Support
- Privacy FAQ for Microsoft Copilot — Microsoft Support
- Improving your recommendations on our apps with AI at Meta — Meta
- Remove posts, chats and media from Meta AI — Meta Help Center
