Skip to content
Techlantis
Guides

How to keep your data private in ChatGPT, Claude, Gemini and Copilot

Most AI chatbots can use your conversations to improve their models unless you change a setting. Here's where each switch lives, how temporary chats work, and what opting out doesn't cover.

10 min · Updated September 25, 2026 · Tools: ChatGPT, Claude, Gemini, Microsoft Copilot, Meta AI

  1. 1ChatGPT: turn off model training. On the web, open Settings, then Data controls, and switch off 'Improve the model for everyone'. On the app: sidebar, profile icon, Data controls. It applies to new conversations.
  2. 2Claude: check 'Help improve our AI models'. Open Settings, then Privacy, and set 'Help improve our AI models' the way you want. Turning it off covers new chats and future training runs.
  3. 3Gemini: turn off Keep Activity. Open Settings & help, then Activity, and turn off Keep Activity (or turn it off and delete). It's on by default for adults and lets Google use chats to improve its models.
  4. 4Use temporary chats for sensitive one-offs. ChatGPT's Temporary chat, Claude's incognito chat (the ghost icon) and Gemini's temporary chat stay out of your history and aren't used for training. Providers may still keep them briefly for safety, from 72 hours to 30 days.
  5. 5Review memory separately. Assistants that remember things about you keep that memory apart from your chat history. Check the memory settings and delete what you don't want kept.
  6. 6Never paste secrets. Don't enter passwords, ID numbers, card or bank details, or health records into a chatbot, even with training switched off.

AI assistants learn from the conversations people have with them. For personal accounts, that often includes yours, unless you change a setting. Each company puts the switch in a different place, so here's where to find it and what it actually does.

The training switches

Assistant Setting Where
ChatGPT Improve the model for everyone Settings › Data controls
Claude Help improve our AI models Settings › Privacy
Gemini Keep Activity Settings & help › Activity

Microsoft Copilot is in transition. Microsoft says its updated Copilot app, released in August 2026, doesn't use your prompts and responses to train its foundation models. Older versions had "Training on conversation activity" settings under Privacy. Check which version you have in Microsoft's privacy controls.

Meta AI: Meta says it uses AI interactions to improve its AI. We couldn't confirm a simple opt-out switch in Meta's official help pages. You can delete your Meta AI chats under Settings › Data & privacy in the app, though Meta's terms note that deleting may not remove every copy.

What opting out doesn't do

  • It isn't retroactive. Anthropic says conversations already used in finished training stay there.
  • Feedback can override it. OpenAI says a thumbs up or down can send that conversation to training anyway.
  • Human review can continue. Google keeps chats already reviewed by people for up to three years, even if you delete your activity.
  • Memory is separate. In ChatGPT, deleting a chat doesn't delete a memory created from it.

Work accounts are different

Business and enterprise plans generally come with stricter defaults. OpenAI doesn't train on ChatGPT Business, Enterprise or Edu content by default, and on Google Workspace your administrator controls Gemini's activity settings. If you use AI at work, ask your IT team which plan you're on before sharing company data.

For picking an assistant in the first place, see how to choose an AI assistant.

Quick questions

Does ChatGPT use my chats to train its models?

For personal accounts it can, unless you turn off 'Improve the model for everyone' in Settings, Data controls. OpenAI says it doesn't train on ChatGPT Business, Enterprise or Edu content by default.

Are temporary chats really private?

They stay out of your history and aren't used for training, but they're not instantly erased: OpenAI may keep them up to 30 days, Anthropic keeps incognito chats 30 days, and Google keeps temporary chats 72 hours.

Can people at AI companies read my chats?

Sometimes. OpenAI and Google both say limited staff or reviewers may see some conversations, for example for safety checks or to improve the service. Google says reviewed chats are disconnected from your account.

Does giving a thumbs up or down affect my privacy?

It can. OpenAI says rating a response can include that whole conversation in training even if you opted out, and Google uses feedback even when Keep Activity is off.

Sources

Settings and prices change. We check guides against these sources and update the date above when we do.