How to Stop ChatGPT, Claude and Gemini Training on Your Data

ChatGPT, Claude and Gemini can all use conversations from personal accounts to improve their AI models unless you switch that off. Each has a single setting that stops it, plus a private chat mode for one-off conversations. This guide shows where to find them, how long each company still keeps your data, and how business plans differ, using OpenAI’s, Anthropic’s and Google’s own help and privacy pages as of 11 November 2026.

Quick comparison: AI training settings at a glance

	ChatGPT (OpenAI)	Claude (Anthropic)	Gemini (Google)
Setting to turn off	“Improve the model for everyone”	“Help improve our AI models”	Keep Activity (Gemini Apps Activity)
Where	Settings → Data Controls	Settings → Privacy	myactivity.google.com/product/gemini
Kept after you opt out	Chats stay in your history	Deleted chats removed from servers within 30 days	Chats still saved for 72 hours
Private chat mode	Temporary Chat, deleted within 30 days	Incognito chat, kept 30 days by default	Temporary chat, kept 72 hours
Business plans trained by default?	No (Business, Enterprise, Edu, API)	No (Claude for Work, API)	No (Workspace accounts)

How to stop ChatGPT training on your data

OpenAI’s setting is called Improve the model for everyone. Turning it off means new conversations still appear in your history but are not used for training.

1. On the web: click your profile icon at the bottom left, open Settings, then Data Controls, and turn off Improve the model for everyone.
2. On iPhone: tap the three dots at the top right, then Settings and Data Controls, and switch it off.
3. On Android: open the menu (three lines, top left), then Settings and Data Controls, and switch it off.
4. Not logged in? Click the question mark icon at the bottom right on the web, then go to Settings and Data Controls.

The choice is tied to your account, so it applies on every device. OpenAI says there are no limits on turning it off.

Temporary Chat is ChatGPT’s private mode. These chats don’t appear in your history, can’t see your memories and are never used for training. OpenAI deletes them from its systems within 30 days and says it reviews them only when needed to check for abuse.

Data Controls is also where you export your data or delete your account. Deleting your account can’t be undone.

Business plans: OpenAI says that by default it does not train on data from ChatGPT Business, ChatGPT Enterprise, ChatGPT Edu or its API. API customers can opt in from the API dashboard, and qualifying organisations can set how long OpenAI keeps their data.

How to stop Claude training on your data

Anthropic’s setting is Help improve our AI models. It covers Claude’s consumer plans (Free, Pro, Max and Ultra), including Claude Code used through those accounts.

1. In Claude on the web or desktop, click your name, then Settings.
2. Select Privacy. You can also go straight to claude.ai/settings/data-privacy-controls.
3. Under Help improve our AI models, turn the toggle off.
4. On mobile, the path is the same: your name, then Settings, then Privacy.

Once it’s off, new chats and coding sessions aren’t used for future training. Anthropic says data already in a training run that has started, or in a model already trained, is not removed.

How long Anthropic keeps your data: if the setting is on, de-identified data from new or resumed chats may stay in training pipelines for up to five years. Deleted chats vanish from your history straight away and are removed from Anthropic’s back-end systems within 30 days. There are exceptions. Chats flagged by safety systems can be kept for up to two years (and their safety scores for up to seven years), and thumbs-up or thumbs-down feedback is kept for five years. Flagged chats may also be used to improve Anthropic’s safety systems even when the setting is off.

Incognito chats: click the ghost icon at the top right of a new chat (it isn’t available inside Projects). Incognito chats aren’t used for training, don’t use or add to Claude’s memory, and are kept for 30 days by default. They are available on all plans. On Team and Enterprise plans, account owners can still see them in data exports.

Business plans: Anthropic says it does not use inputs or outputs from commercial products, such as Claude for Work and its API, for training by default. The exception is conversations you send feedback on, or where you choose to allow it.

How to stop Gemini training on your data

Google’s control is the Keep Activity setting, part of Gemini Apps Activity. You manage it at myactivity.google.com/product/gemini while signed in to your Google Account.

When Keep Activity is off, Google says future chats won’t show in your activity and won’t be used to train its AI models, unless you choose to send feedback. Google still saves chats for 72 hours so Gemini can reply, process feedback and keep the service safe.

• Auto-delete: if you leave activity on, Google deletes it after 18 months by default. You can change that to 3 months or 36 months, or turn auto-delete off.
• Human review: chats reviewed by people are disconnected from your account first and kept for up to three years. Deleting your activity does not delete chats that have already been reviewed.

Temporary chat: on gemini.google.com, click the temporary chat icon next to New chat. These chats don’t appear in your activity or recent chats, aren’t used for training and are kept for 72 hours. Temporary chats turn off Gems, personalisation and some connected apps, and they only work with personal Google Accounts.

Work and school accounts: Google says Workspace does not use customer data to train models without the customer’s permission, and that Gemini app chats on these accounts aren’t reviewed by humans. Since June 2026, Workspace admins can decide whether users may use temporary chats and delete conversations, and any Google Vault retention rules still apply.

Pros and cons of opting out

ChatGPT: opting out is one switch and keeps your full chat history. The downside is that your saved chats still stay on OpenAI’s servers until you delete them.

Claude: the setting is clear and incognito works on every plan. But leaving the setting on means data may be kept for up to five years, and safety-flagged chats have their own retention periods.

Gemini: temporary chats are kept for the shortest time (72 hours). But Keep Activity mixes training with your chat history, and human-reviewed chats are kept for up to three years.

Which option should you use?

• You want history but no training: turn off the training setting in ChatGPT or Claude. In Gemini, turning off Keep Activity also stops your history, so use temporary chats for sensitive questions instead.
• You are sharing something sensitive once: use Temporary Chat, incognito chat or a Gemini temporary chat, and avoid pressing thumbs up or down, because feedback can be kept longer.
• You work with client or company data: use a business plan (ChatGPT Business or Enterprise, Claude Team or Enterprise, Google Workspace), where training is off by default.

If you are still choosing a paid plan, see our ChatGPT Plus vs Claude Pro vs Gemini comparison and the latest prices on our AI subscription price tracker.

Frequently asked questions

Does paying for ChatGPT Plus or Claude Pro stop training automatically?

No. Personal paid plans such as ChatGPT Plus, Claude Pro, Max and Ultra use the same training setting as the free plans, so you have to turn it off yourself. Business and enterprise plans are not used for training by default.

Does opting out remove data that was already used?

No. Anthropic says data already in a training run or a trained model is not removed. OpenAI says the setting applies to new conversations. Delete old chats you don’t want stored.

Are temporary or incognito chats completely private?

They aren’t used for training, but they are not deleted immediately. ChatGPT removes Temporary Chats within 30 days, Claude keeps incognito chats for 30 days by default, and Gemini keeps temporary chats for 72 hours, mainly for safety checks.