About 900 million people use ChatGPT every week — and you can stop OpenAI from using your future chats to train its models. OpenAI provides built-in controls to opt out of model training, delete past conversations, and submit formal privacy requests; follow five practical steps to see what the service holds about you and reclaim control over your data.

Start with the training opt-out

Stop OpenAI from using your future conversations for model training by opening Settings → Data controls and toggling off "Improve the model for everyone." Click Done to save the change. This prevents future chats from being added to training datasets but does not remove content already used to train models.

Delete old chats and clean history

To remove past conversations, either delete individual threads from the left-hand sidebar (use the three-dot menu beside a chat) or clear everything via Settings → Data controls → Delete all chats. Deleted chats disappear from your history immediately, but OpenAI says permanent removal from its systems may take up to 30 days.

Use the privacy portal to make formal requests

OpenAI’s privacy portal lets you submit formal requests—choose the option for a consumer ChatGPT account and select "Don't train on my content." You may be asked to sign in. Filing a request there creates a formal record beyond toggling in-app settings.

Limit what you share going forward

Avoid pasting sensitive details such as financial numbers, Social Insurance Numbers, passwords, or detailed medical records into chats. Prefer general descriptions instead of names and specific dates. Quick habits to follow:

  • Replace exact identifiers with placeholders (e.g., CLIENT_NAME).
  • Use on-device or local tools for sensitive tasks when possible.
  • Ask if a non-cloud solution can complete the job before sharing personal data.

Create an audit log for yourself

Keep a short log of when and why you used ChatGPT for tasks involving user data, and note which chats should be preserved or deleted. For teams, adopt a simple policy about what can be pasted into shared chatbot accounts to reduce accidental exposure.

What to expect after you take action

Turning off the model-improvement toggle stops OpenAI from using your future content for training, and deleting conversations removes them from your view while the company processes permanent removal (up to 30 days).

Related Articles

OpenAI says deleted chats can take up to 30 days to be permanently removed from its systems.

This article was created with AI assistance.