How to Delete Your Data From AI Chatbots, and Your Rights

To delete your data from an AI chatbot, delete the conversations in the app, turn off chat history or training in its settings, and delete your account if you want everything gone. Deleting a chat removes it from your history right away, but companies keep copies on their back-end systems for a period, longer if a chat was flagged or held for legal reasons. If you live in the EU, the UK, California or another state with a privacy law, you also have a legal right to ask what a company holds about you and to have it erased.

Here’s how that works for the big three, and what the law gives you.

What does deleting an AI chat actually remove? #

Three copies of your data can exist, and the delete button reaches them differently.

Chats reviewed by people or flagged by safety systems usually follow separate, longer schedules, and a court order can force a company to preserve data it would otherwise delete. That happened to OpenAI in 2025 during the New York Times lawsuit, as we cover in do humans read your AI chats.

How to delete your ChatGPT data #

In ChatGPT’s settings you’ll find controls to export your data, delete all chats, turn off “Improve the model for everyone” and delete your account. OpenAI has moved these between menus over time, so check under both Data Controls and Account. OpenAI also handles privacy requests, including deletion and access requests, through its privacy portal at privacy.openai.com.

Before you delete your account, export your data if you want a copy of your conversations. Account deletion is permanent. For future chats you don’t want kept in your history at all, use Temporary Chat.

How to delete your Gemini activity #

Google’s Gemini activity help page gives these steps:

1. Go to gemini.google.com.
2. Open Settings & help, then Activity.
3. Click Delete and choose a range: the last hour or day, a specific period, a specific day, or All time.

To stop saving new chats, click the On toggle near the top of the same page and choose Turn off or Turn off and delete activity. Google notes that “even when Keep Activity is off, your conversations will be saved with your account for up to 72 hours.” You can also set auto-delete to 3, 18 or 36 months. Google says chats that were reviewed by people are kept for up to three years even if you delete your activity.

How to delete your Claude chats #

Anthropic’s data retention page says a deleted conversation is “removed from your chat history immediately” and “deleted from our back-end storage systems within 30 days.”

The model-training setting matters more than the delete button:

• Training setting on: Anthropic may keep chats “in a de-identified format for up to 5 years in our model training pipelines.”
• Training setting off: Anthropic says it “will not use your previous or new chats or coding sessions for future model training.”
• Flagged for policy violations: inputs and outputs can be kept for up to 2 years, and safety classification scores for up to 7 years.

So if you care about deletion, turn the training setting off first, then delete the chats you don’t want kept.

What are your legal rights over AI chat data? #

AI companies are bound by the same privacy laws as any other business that holds your personal data.

In the EU and UK: the GDPR #

Under Article 17 of the GDPR, you have the right to have your personal data erased “without undue delay” when, among other grounds, it’s no longer needed, you withdraw consent, or you object to its processing. You also have rights to access your data, correct it and take a copy with you.

Article 12 sets the clock: the company must respond “within one month of receipt of the request,” extendable by two more months for complex cases if it tells you why. Requests are free unless they’re clearly unfounded or excessive.

The right isn’t absolute. A company can keep data it’s legally required to keep, or needs for legal claims.

In California: the CCPA #

The California Privacy Protection Agency lists six rights: to know what’s collected about you, to delete it, to correct it, to opt out of its sale or sharing, to limit the use of sensitive personal information, and to equal treatment for using these rights. Businesses must confirm your request within 10 business days and respond within 45 calendar days. You can make a right-to-know request up to twice a year for free.

Several other U.S. states have passed comprehensive privacy laws with similar access and deletion rights. Even where no law applies, the major AI companies accept deletion requests from all users.

How to send a data deletion request #

1. Find the company’s privacy request page. Look for “privacy request,” “your privacy choices” or “data subject request” in its privacy policy or help center.
2. Ask for a copy first, if you want to see what they hold.
3. Request deletion, naming the account email you used.
4. Say which law applies, such as the GDPR or the CCPA, if you know it.
5. Keep a record of the date and any confirmation, and follow up if the deadline passes.

A short request works fine:

I request deletion of all personal data you hold about me, including chat history, account information and any copies retained for model training, under [GDPR Article 17 / the CCPA]. My account email is [email]. Please confirm when this is complete.

If a company in the EU ignores you, you can complain to your country’s data protection authority. In California, you can file a complaint with the CPPA.

Can your data be removed from a trained AI model? #

Not in any practical sense. Turning off training and deleting chats stops your data from being used in future training. It doesn’t reach into a model that already learned from it. That’s the strongest argument for turning training off before you share anything personal, rather than cleaning up afterwards. Our guide to keeping personal data out of AI chatbots covers what to leave out in the first place.

The simplest option: data that never leaves your phone #

Every step above exists because a company holds a copy of your conversations. An AI that runs on your phone skips the whole problem.

Personal LLM runs open models on your phone’s own chip, with no account and no server. Your chats are stored on the phone, so deleting one in the app deletes it, and uninstalling the app removes everything. There’s no retention window, no training pipeline and no one to send a request to. The one copy to remember is any backup you choose to export: the app can save every chat as a JSON file, and that file is yours to keep or delete.

It’s a smaller model than ChatGPT, Gemini or Claude, so it suits personal writing, questions about your own documents and private conversations better than questions that need the latest facts. See is it safe to tell AI personal things for which topics are worth keeping local.

Frequently asked questions #

Does deleting my ChatGPT account delete all my data? #

It deletes your account and starts the removal of your data from OpenAI’s systems, subject to retention for safety, legal and security reasons. It can’t remove anything already used to train a model. Export your data first if you want a copy.

How long do AI companies keep deleted chats? #

Anthropic says deleted Claude chats leave its back-end within 30 days. Google keeps Gemini chats for up to 72 hours even with activity off, and reviewed chats for up to three years. Legal holds and safety flags can extend any of these.

Can I make an AI company delete my data if I’m not in the EU or California? #

Often, yes. Several U.S. states have their own privacy laws, and the major AI companies accept deletion requests from all users through their privacy portals or settings, whether or not a law requires it.

Does deleting a chat stop it being used for training? #

Not by itself if training is already on and the chat has been collected. Turn off the training setting first. Anthropic, for example, says that with the setting off it won’t use your previous or new chats for future training.

How do I find out what data an AI company has on me? #

Use its data export tool, or send an access request through its privacy page. Under the GDPR the company must respond within one month. Under the CCPA it has 45 calendar days.