How to Stop AI Chatbots From Training on Your Chats -

Quick answer: Most big AI chatbots let you stop your chats being used for training with one setting. In ChatGPT, turn off “Improve the model for everyone”; in Gemini, turn off Keep Activity; in Claude, turn off “Help improve our AI models”; in Microsoft Copilot, turn off “Training on conversation activity” and “Training on voice conversations.” Meta AI has no simple equivalent switch, so the safest option there is to share less.

When you chat with an AI assistant, what you type can be stored and, depending on your settings, used to improve future models. Some services also have people review a sample of conversations. That’s worth knowing before you paste in a medical report, a work contract or your child’s school details.

Below are the exact steps for each major chatbot, as of October 2026, plus what each setting does and doesn’t cover. Menus move around, so if a label looks slightly different in your app, look for the privacy or data controls section.

Why do AI chatbots train on your conversations?

AI companies use real conversations to spot mistakes and make their models better at everyday tasks. Google is unusually direct about this. Its Gemini Apps Privacy Hub says that with activity saving on, Google uses your data “to provide, develop, and improve its services (including training generative AI models),” and that a subset of chats is checked by human reviewers.

Google’s advice is good advice for every chatbot: “don’t enter data that’s confidential or that you wouldn’t want a reviewer to see.” Turning off training reduces how your chats are used, but it doesn’t make an AI chat as private as a note on your own phone.

How to stop ChatGPT from training on your chats

OpenAI’s setting is called “Improve the model for everyone.” According to its data controls help page, it’s available to signed-in Free, Go, Plus and Pro users.

On the web or desktop

Click your profile name or icon and open Settings.
Go to Data controls.
Turn off Improve the model for everyone.

On the iPhone or Android app

Open the sidebar and tap your profile icon to open Settings.
Tap Data controls.
Turn off Improve the model for everyone.

Turning this off “does not delete or hide saved chats,” OpenAI says. Your history stays as it is. For one-off sensitive questions, use Temporary Chat: those chats don’t appear in your history, though OpenAI says they “may be retained for up to 30 days for safety purposes.”

In September 2026 OpenAI also added a Privacy Center to ChatGPT that gathers chat privacy, memory, data use and account security settings in one place, according to its release notes. If you’d also like ChatGPT to stop remembering details about you, that’s a separate setting from training.

How to stop Google Gemini from using your chats

Gemini’s control is called Keep Activity (you may still see it referred to as Gemini Apps Activity). Google’s help page gives these steps:

Go to gemini.google.com (or open the Gemini app).
Open Settings & help, then Activity.
Near the top, click On, then choose Turn off or Turn off and delete activity.

There are trade-offs. Even with Keep Activity off, Google says your conversations are “saved with your account for up to 72 hours” so it can run the service. Chats you have with Gems won’t be saved, though the Gems you create are kept.

One more detail matters: chats that were already picked for human review are kept “for up to three years,” even if you later delete your activity, because Google disconnects them from your account. Turning the setting off stops future chats from being reviewed for model improvement, not past ones.

How to stop Claude from training on your chats

Anthropic’s setting covers Claude Free, Pro and Max accounts, according to its privacy help page.

Select your name, then Settings (on desktop you can also go straight to claude.ai/settings/data-privacy-controls).
Choose Privacy.
Turn off Help improve our AI models.

The steps are the same in the mobile app. Anthropic says that after you switch it off, it won’t use new chats and coding sessions for future training. Two limits apply: data already in training that has started stays there, and conversations flagged by its safety systems may still be used for trust and safety work.

How to turn off model training in Microsoft Copilot

Copilot has two separate switches, one for text and one for voice. Microsoft’s privacy controls page lists the paths:

copilot.com: profile icon > your profile name > Privacy.
Copilot app for Windows or Mac: profile icon > Settings > Privacy.
Mobile app: menu > profile icon > Account > Privacy.

In each case, turn off Training on conversation activity and Training on voice conversations. Microsoft says opting out excludes your future conversations from training.

What about Meta AI on WhatsApp, Instagram and Facebook?

Meta AI works differently. There isn’t a single “don’t train on my chats” toggle like the ones above. Meta also uses your conversations with Meta AI to personalise what you see: from 16 December 2025, Meta said it would use those chats to tailor content and ads across its apps “in most regions,” according to its announcement.

Meta says it doesn’t use sensitive topics, such as health, religious or political views, to show ads. You can adjust ads and feed recommendations through Ads Preferences, but that isn’t the same as stopping the data being used.

Our advice: treat Meta AI as public-ish. Use it for general questions and avoid personal details. Your normal WhatsApp chats with friends and family are a separate matter from chats with the Meta AI assistant.

Quick comparison: AI training settings by chatbot

Chatbot	Setting to turn off	Private mode
ChatGPT	Improve the model for everyone	Temporary Chat (kept up to 30 days)
Gemini	Keep Activity	Chats kept up to 72 hours when Keep Activity is off
Claude	Help improve our AI models	Check your Claude app for current options
Copilot	Training on conversation activity; Training on voice conversations	Not covered here
Meta AI	No single training toggle	Share less; manage Ads Preferences

Habits that protect you whatever the settings

Don’t paste passwords, card numbers, Aadhaar or Social Security numbers, or full medical records into any chatbot.
Remove names and account numbers from documents before asking for a summary.
Use a private or temporary mode for one-off sensitive questions.
Delete old chats you no longer need.
Check the settings again after big app updates, since new features sometimes come with new switches.

Privacy also matters when the content comes to you. Our guide on how to spot AI-generated images and deepfakes covers the other side of staying safe with AI. For more guides, see our AI section.

Frequently asked questions

Does turning off training delete my old ChatGPT chats?

No. OpenAI says turning off “Improve the model for everyone” doesn’t delete or hide saved chats. You have to delete chats separately if you want them gone.

Do I lose features if I turn off Gemini Keep Activity?

Your chats won’t be saved to your history beyond the 72-hour window, and chats with Gems won’t be saved. You can still use Gemini normally for new questions.

Can I stop Meta AI from using my chats?

There’s no single setting that does this outside of regional legal rights. The practical approach is to avoid sharing personal details with Meta AI and to manage your ad preferences.

Is a “temporary” or “incognito” chat completely private?

Not completely. For example, OpenAI says Temporary Chats may be kept for up to 30 days for safety. These modes keep chats out of your regular history, but the company can still hold them for a while.

How to Stop AI Chatbots From Training on Your Chats