Turn Off AI Training on Your Chats: 3 Settings to Change Today

WatchInstagram

To stop ChatGPT, Claude and Gemini using your chats to train their models, switch off one setting in each app: Improve the model for everyone in ChatGPT, the model improvement toggle in Claude, and Keep Activity in Gemini. Then use a temporary chat (ChatGPT, Gemini) or an incognito chat (Claude) for anything sensitive.

The whole fix takes about three minutes. Turning these off doesn't delete anything from your history, it only stops your new chats being used for training.

Below are the menu paths, what each setting does and doesn't cover, and every official link. It's written for personal accounts.

Which settings stop AI apps training on your chats?

Each app has one switch that decides whether your new chats can be used to train its models, plus a private mode that sits outside it.

ChatGPT: stop training with Improve the model for everyone

Turning this off means OpenAI won't use your new conversations to train its models. Your chats stay in your history, because the setting doesn't delete or hide anything. A temporary chat isn't used to improve OpenAI's models either, unless you save it, but OpenAI says it may keep a copy for up to 30 days for safety.

Claude: stop training with the model improvement toggle

Turning this off means Anthropic won't use your new chats and coding sessions for future training. Anything already in a training run or a trained model stays there, Anthropic says. Incognito chats aren't used for training either, and Anthropic keeps them for 30 days by default.

Gemini: stop training by turning off Keep Activity

With Keep Activity off, Google says your future chats won't appear in your Activity or be used to train its AI models, unless you send feedback. They're still saved for 72 hours so Gemini can respond and Google can protect its service. Temporary chats aren't used for training either, and Google keeps those for 72 hours too.

What don't these settings cover?

They cover training, not everything else a company does with your chats.

Safety checks still run. Anthropic says that if its systems flag a conversation, it can still be used to detect harmful content and enforce its policies, even if you've opted out. Google says a subset of chats are reviewed by human reviewers, and that it still uses chats to respond to you and protect its users with Keep Activity off or in a temporary chat.

Feedback buttons can undo your choice. OpenAI says the entire conversation behind a thumbs up or down may be used to train its models, even if you've opted out. Anthropic says it stores the whole conversation and may use it for training. Google says that with Keep Activity off, feedback brings in the last 24 hours of your chats.

How do you turn these settings off?

Menu names change, so if yours differ, follow the linked help page for that app.

1. ChatGPT

On the web, open your account menu, go to Settings → Data controls, switch off Improve the model for everyone and select Done. In the mobile app, open the sidebar, tap your profile icon, then go to Settings → Data controls.

For a temporary chat, open a new chat and select Temporary. Choose Unpersonalized before your first message if you don't want it using your memories.

2. Claude

Click your name, then go to Settings → Privacy and switch off the model improvement toggle. Anthropic's own steps call it "Help improve our AI models", and the label in your app may read differently.

For an incognito chat, click the ghost icon in the top right of a new chat. It doesn't show inside projects.

3. Gemini

Go to gemini.google.com, click Settings & help → Activity, and turn off Keep Activity near the top. Google says it's on by default if you're 18 or over.

You can delete earlier activity from the same page, but Google says chats a reviewer has already seen aren't removed that way. For a temporary chat, click Temporary chat next to New chat.

What should you never paste into an AI chat?

Here's the rule from the video: if you wouldn't send it in an email to a stranger, don't paste it into a chat.

Client data, passwords, health details and legal details all fail that test. Anthropic's own help page tells people to be thoughtful about financial details, health records, passwords and confidential documents.

Do these steps work on a work account?

Not the same way. Your admin sets the rules on work and team plans, so this page is for personal accounts.

OpenAI says it doesn't use content from ChatGPT Business, Enterprise or Edu workspaces for training by default, and Anthropic says the same for its commercial products. Google says Gemini activity settings on work or school accounts are controlled by your Workspace administrator, and temporary chats aren't available on them.

This is general information, not legal advice.

Want more on using AI safely in a real business? Read our insights on AI implementation, or take the free AI bottleneck test to see where AI would save your team the most time.