I asked six AI engines the same question. Who keeps my data?
The answers aren't the same, and the differences matter if you paste anything about a client into a chat box. Here's my cheat sheet.
ChatGPT
On the Free, Plus and Pro plans, OpenAI says it may use your conversations to train its models unless you turn it off. The switch is under Settings, Data controls, Improve the model for everyone. I did a whole episode on that one switch.
Claude
Anthropic flips the default. Claude uses your chats for training only if you choose to allow it. Incognito chats are not used to improve Claude, even if you've turned model improvement on.
One honest footnote. Anthropic's page also lists chats flagged for a safety review as something it may use, to train the systems that catch misuse. So "never" has an asterisk.
Gemini
Gemini is different. With Keep Activity on, Google says it uses your activity to improve its services, including training generative AI models, with the help of human reviewers. Chats that a reviewer looks at are disconnected from your account and kept for up to three years. Deleting your activity doesn't delete those.
If you want to stop future chats from being reviewed, Google says to turn Keep Activity off.
Grok and Meta AI
I couldn't confirm their pages well enough to put a claim on screen. So check the data settings in your own account before you trust either one with anything private.
Your own machine
The private option is an open model running on your own computer. Nothing leaves it. It takes a little setup, and the models are a step behind the big ones, but for sensitive work that trade is worth it.
Not sure where you stand?
Take the free AI readiness quiz: quiz.techtony.co/ai-readiness
Sources (checked October 20, 2026. These pages change, so check them before you rely on them):
ChatGPT, Claude, Gemini, Grok and Meta AI are trademarks of their respective owners. This post is not affiliated with or endorsed by any of them.