ChatGPT's data controls, explained
What OpenAI does with your chats by default, and the settings that change it.
Updated 31 Jul 2026 fast-moving — check a current source
Behaviour per OpenAI's data-controls documentation, checked late July 2026. Consumer settings change more often than enterprise commitments — verify before relying on either.
The single most common question about ChatGPT at work is “does it train on what I type?” The answer depends on which product you’re using and which settings are on — so it’s worth knowing the specifics rather than assuming either the best or worst case.
Consumer vs business
In the consumer app, whether your conversations may be used to improve models depends on your settings, and there’s a control to turn that off. In the business and enterprise products (Team, Enterprise, and the API), business data is not used to train models by default — a different commitment aimed at organisations.
The controls worth checking
- Improve the model for everyone — the toggle that governs whether your chats can be used in training. Off means they’re not.
- Temporary chat — a session that isn’t saved to your history.
- Data retention — how long conversations are kept, which differs by product and configuration.
“Not used for training” is not the same as “never stored”. Chats may still be retained for a period to run the service and handle abuse — check the retention terms, not just the training toggle.
For sensitive work, treat the principle the same way as with Copilot’s boundary: know the setting, don’t assume it.
Sources
Everything above was checked against these on 31 Jul 2026. Providers change things without notice — if a detail matters to a decision, follow the link.