Fanetti AI
Menu

Keep your personal info out of AI chats

Updated 3 min read

New to AI

Part of: Your first week with AI · Step 3 of 6

Settings and links checked October 22, 2026 against each company’s own help pages.

What you type into an AI chat may be stored, and many apps can use your chats to improve their models unless you change a setting. The simplest habit: don’t paste anything you wouldn’t want a stranger to read.

Don’t paste

  • passwords, PINs and security codes
  • bank, card or government ID numbers
  • medical details, yours or anyone else’s
  • other people’s names, addresses or photos without their OK
  • confidential work documents, unless your employer has approved the tool

Need help with a document? Remove names and numbers first, or replace them with placeholders like [name] and [account number].

Turn off training on your chats

Each company explains its own settings. Here’s what their help pages said when we checked on October 22, 2026. Menus move around, so if the app looks different, follow the official page.

  • ChatGPT: in Settings, open Data controls and turn off Improve the model for everyone. OpenAI says this is on by default for personal Free, Plus and Pro accounts. Temporary chats aren’t used to train its models, though they may be kept for up to 30 days for safety. (Data controls in ChatGPT)
  • Claude: Anthropic uses chats from its consumer plans to improve its models if you choose to allow it in your privacy settings. Incognito chats aren’t used, even if that setting is on. (Is my data used for model training?)
  • Gemini: turn off the Keep Activity setting to stop your future chats being reviewed and used to improve Google’s services. Google says it still keeps chats for 72 hours to respond to you and keep the service safe. Temporary chats aren’t used to train its models. (Gemini Apps Privacy Hub)
  • Grok: in the app, go to Settings, then Data Controls, and turn off Improve the model. On grok.com, it’s under Settings, then Data. Private Chat (the ghost icon) isn’t used for training. (Grok consumer FAQ)
  • Mistral Vibe (formerly Le Chat): Mistral says users aren’t opted out by default. In the mobile app, open Settings, then Data & Account Controls, and untick Enable data sharing; on the web, it’s a privacy toggle in the admin panel. (Opting out of training)

One catch with feedback buttons: OpenAI, Anthropic and xAI all say that if you rate an answer with thumbs up or down, that conversation may be used for training even if you’ve opted out. Google says the same applies when you send feedback with Keep Activity off.

Turning off training isn’t deleting

Turning off training doesn’t remove chats you’ve already had. Delete conversations you no longer need, and check what happens to your data if you close your account.

Private by design

If you want nothing to leave your computer, run AI on your own computer. For more safety habits, read Using free AI models safely, and see the free options in Resources.

The short version

Leave private details out, turn off training in the app’s settings, skip the feedback buttons on anything sensitive, and use local AI when it really matters.