How to stop ChatGPT, Claude, Gemini, Copilot and Grok training on your chats
In short: Each of the big five chatbots has a setting that stops your new chats being used to train its AI. Turn it off once on each account you use, and remember that feedback buttons and some safety checks can still send chats to the company.
Last checked: · Next check: · Settings and menus change often.
General information only, not legal advice. Laws and rules differ between countries and regions, so always check what applies where you live, and get professional advice for your situation.
At a glance
| Chatbot | Setting to turn off | Where to find it | Official help page |
|---|---|---|---|
| ChatGPT | "Improve the model for everyone" | Web: account menu > Settings > Data controls. App: sidebar > profile icon > Data controls | OpenAI: Data controls |
| Claude | "Help improve our AI models" (under Privacy) | Web, desktop and app: your name > Settings > Privacy | Anthropic: model improvement settings |
| Gemini | "Keep Activity" | Computer: gemini.google.com > Settings & help > Activity > Turn off. Phone app: we couldn't confirm the current steps; check the help page | Google: Manage your Gemini Apps activity |
| Copilot | "Training on conversation activity" and "Training on voice conversations" | copilot.com: profile icon > your profile name > Privacy. Windows/macOS app: profile icon > Settings > Privacy. Phone app: menu > profile icon > Account > Privacy | Microsoft: Copilot privacy controls |
| Grok | "Improve the model" | App: Settings > Data Controls. grok.com: Settings > Data | Grok consumer FAQ |
Menus change often. If what you see doesn't match, use the official help page for that chatbot.
What turning it off does (and doesn't do)
- ChatGPT: new chats won't be used for training, and your saved chats stay in your history. If you press thumbs up or down, the whole conversation may still be used for training. Temporary chats aren't used for training but may be kept for up to 30 days for safety.
- Claude: new chats won't be used for future training. Chats flagged by safety systems may still be used for safety work.
- Gemini: your future chats won't be used to train Google's AI unless you send feedback. They're still kept for 72 hours. Chats already checked by human reviewers are kept for up to three years, even if you delete your activity.
- Copilot: your future conversations won't be used to train its AI models. Microsoft says they can still be used for product improvement, advertising, safety and security. Some account types work differently; check Microsoft's Copilot privacy FAQ.
- Grok: only your new chats are excluded. Feedback you choose to give may still be used. Grok used inside X has its own setting in X. In some regions, if you use Grok without signing in, you can't opt out.
Checklist
- Turn the training setting off in each chatbot you use, on each account (personal and any others).
- Sign in before you chat, so your choice is saved to your account.
- Think before pressing thumbs up or down on a chat with private details in it.
- Use a temporary or private chat for one-off sensitive questions: ChatGPT Temporary chat, Gemini temporary chat, or Grok Private Chat.
- Delete old chats you don't need. Turning training off doesn't delete anything.
- Put a reminder in your calendar to check again in a month, because settings change.
Common questions
Does turning training off delete my old chats?
No. In ChatGPT, Claude, Gemini, Copilot and Grok, turning training off affects future chats. Delete old chats separately if you want them gone.
Is a paid plan private by default?
Not always. Personal paid plans usually follow the same settings as free plans. OpenAI says business workspaces aren't used for training by default, and Grok's FAQ says the same for its business customers.
Do I need to do this on my phone and computer?
For ChatGPT, the setting applies across your devices when you're signed in. For the others, check each app you use.
Guide only. No affiliate or sponsored links on this page. Disclaimer · Affiliate disclosure