ChatGPT is recording your conversations: take back control in 5 minutes

By default, ChatGPT keeps your conversations in your history and, on a consumer account, may use them to improve its models. Good news: three settings, reachable in five minutes, sharply reduce that exposure. Less good news: those settings act on the provider's side and don't cover everything. The only protection that depends on no setting is not putting the sensitive data in the conversation in the first place.

Why your conversations are kept by default

An assistant like ChatGPT keeps a history to serve your exchanges back to you, and — on consumer plans — the provider may use those conversations to train its models unless you turn it off. It's an experience choice, not a fate: the official documentation (OpenAI's “Data Controls FAQ”) describes the options to limit this retention and use. You just have to go find them.

The 5-minute fix

Three actions, in the app's settings, cover the essentials:

1. 1Turn off training: Profile → Settings → Data Controls → disable “Improve the model for everyone.” Your future conversations stop feeding training.
2. 2Clear history: in Settings → General → “Delete all chats.” This empties your history (without removing what already fed training).
3. 3Use temporary chat for sensitive exchanges: it doesn't appear in history and doesn't feed memory.

What these settings don't cover

Turning off training doesn't erase what was already used, and above all doesn't make your messages invisible: the text is still transmitted, processed, and may be retained for a time for safety — even when you've disabled training. A setting reduces a specific use; it doesn't remove the fact that the data leaves your device.

The protection that depends on no setting

If every setting leaves a gray zone, the only solid guarantee is upstream: don't hand over the sensitive data. What never leaves your browser can't be retained, reused or reviewed — whatever the state of your checkboxes.

• Do the 3 settings: it's good hygiene.
• But don't rely on them for truly sensitive data.
• Remove identities, identifiers and secrets before sending.

That's what ONYRI Sanitize is for: the engine detects sensitive data and replaces it with reversible tokens before sending; detection and the token↔value mapping stay in your browser, and only anonymized text reaches the AI. Your settings stay useful — but they now only apply to already-neutralized content.