Free guide · Oct 7, 2026
Keep your AI chats private
The privacy settings to change in Claude, ChatGPT and Gemini, what gets flagged for human review, and how to run AI on your own computer.
The short version
Three things decide how private your AI chats are:
1. Training settings. Each app can use your chats to train future models. You can turn that off in under a minute.
2. Safety review. Every major AI company scans chats for serious harm. Turning off training does not turn this off. If a chat is flagged, a person can read it, and in rare cases of a credible threat to others, the company can report it to police.
3. Where the model runs. A cloud chatbot keeps your chats on someone else's servers. A local model on your own computer keeps them on your machine.
My rule: treat cloud AI like email, not a diary. Write what you'd be fine with a stranger reading. For anything you'd never want read, run a model on your own computer.
What's in the full guide
- 01Step 1: the two-minute fix
- 02Claude
- 03ChatGPT
- 04Gemini
- 05What gets flagged, and what happens next
- 06Run AI on your own computer
- 07The honest limits