Stop chatbot training on your data with these settings

Most major chatbot platforms use your data for model training by default unless you turn that off yourself. As CNET notes, that matters even more as people increasingly use these tools for sensitive conversations, including therapy, companionship, and health questions.

In most cases, stopping future training requires two separate actions: disabling the setting that allows training, and deleting existing conversations already stored on the service. CNET warns there is no undo button if your data has already been used to train a model.

How to turn off training in major chatbots

For Google, CNET says users need to change settings in both their Google account and Gemini:

• Go to myactivity.google.com
• Open Web & App Activity and turn it off
• Delete existing activity from the same page
• In Gemini, open Settings > Activity and set it to Off
• Manually delete saved Gemini conversations

For Claude, CNET says the path is simpler:

Recommended reading

Proton CTO says trust, not features, is the product

• Log in and open Settings > Privacy
• Turn off Help improve our AI models
• Optionally disable Location metadata
• Delete chats from the Chats sidebar

According to Anthropic’s privacy policy, deleted chats are stored for up to 30 days. If training was previously allowed, turning it off stops use of new and existing chats going forward, but not training already in progress or models already trained.

For ChatGPT, users can go to Settings > Data controls and switch off Improve the model for everyone. CNET says that also disables the related audio and video recording training options. A separate button in the same menu deletes all chats. OpenAI also stores deleted chats for up to 30 days.

Meta offers no guaranteed opt-out

CNET says Meta does not provide a clear, guaranteed way to stop AI training on user data. The available option is to submit a form that Meta may act on. The article says stronger protection comes from making profiles private, limiting posts, or deleting Instagram and Facebook accounts entirely. CNET also says Meta did not respond to its request for comment asking for a guaranteed opt-out method.

For Grok on X, the setting is available under Settings and privacy > Privacy and safety > Grok & Third-Party Collaborators. There, users can uncheck:

• Allow your public data as well as your interactions, inputs, and results with Grok and XAI to be used for training and fine-tuning
• Allow X to personalize your experience with Grok
• Allow Grok to remember your conversation history

That same menu also lets users delete Grok conversation history.

The broader takeaway from CNET’s guide is straightforward: if a chatbot offers a training opt-out, it will usually be buried somewhere in data, privacy, or activity settings.