How to remove personal information from ChatGPT, Gemini, Meta AI, and Claude

The Morse teamPublished November 5, 20269 min read

To remove personal information from ChatGPT, open OpenAI's Privacy Portal at privacy.openai.com, select Make a Privacy Request, then Remove personal data from ChatGPT responses, and send links to the chats that show your information with a reason it should go. Google, Meta, and Anthropic take similar requests for Gemini, Meta AI, and Claude, and as of November 2026 each company describes the request in terms of what its chatbot says about you, with none offering to remove you from a trained model.

We read each company's own help, privacy, and form pages on November 5, 2026, along with the GDPR on EUR-Lex and the California Attorney General's CCPA guidance. This page is about information about you that a chatbot repeats: your name, your address, an old news story, a post you wrote years ago. To clear your own chats, use our guide to deleting ChatGPT history instead.

Every removal form, side by side

Built from each company's pages as read on November 5, 2026.

How do you remove your personal data from ChatGPT?

Through the Privacy Portal, in four steps from OpenAI's help article:

1. Go to privacy.openai.com.
2. Click Make a Privacy Request.
3. Select Remove personal data from ChatGPT responses.
4. Fill in the details and click Submit Request.

Gather the evidence before you start. OpenAI asks for "links to chats, and URLs referenced in these chats," the type of data (a name, email, phone number, or URL), a reason the information is "inaccurate, excessive, irrelevant, or no longer appropriate," and "a government-issued ID or alternative proof." So ask ChatGPT about yourself first, keep the link to each chat where your details appear, and note any web pages it cites.

OpenAI's privacy policy gives a second door for errors. If ChatGPT says something false about you, you can ask for "a correction or removal" through the same portal or by email to dsar@openai.com, and OpenAI says it will consider it "based on applicable law and the technical capabilities of our models."

Will OpenAI always remove it?

No. OpenAI says these rights "are not absolute" and that it may decline "if we have a lawful reason." Besides the public-interest test we cover in what machine unlearning can and can't do, OpenAI weighs whether the information comes from official sources, whether it involves a criminal conviction or allegation, and whether you were a child at the time.

OpenAI says it "cannot independently investigate disputed facts," so a claim that something is wrong needs proof attached, such as a court decision. And information tied to "a person's public or professional activities may be more likely to remain relevant," so a request about a job you hold today is harder to win than one about a private matter from years ago.

Why does ChatGPT know about me at all?

Two ways, and the form handles them differently. OpenAI says ChatGPT may show information about people whose details appear "multiple times on the public internet," because it learned language from that internet. It may also "search the web and cite sources," which puts whatever a page says about you into the answer.

When the answer comes from a web page, OpenAI's form leaves that page online, so ask the site's owner to take it down or use Results about you to drop it from Google Search.

How do you get Gemini to stop showing your information?

File a legal request with Google. The Gemini Apps Privacy Hub, last updated October 26, 2026, says that under certain privacy laws, including the GDPR, you may have the right to "object to the processing of your personal data" or to "ask for inaccurate personal data in Gemini Apps' responses to be corrected" (Google). There are two ways in:

• From a reply: below the Gemini response that shows your information, select More, then Report legal issue.
• From the Help Center: use Create a request in the hub's section "How can I object to the processing of my data or ask for inaccurate data in Gemini Apps' responses to be corrected?", which opens Google's legal removal tool.

The same hub warns that Gemini Apps "may produce inaccurate or offensive information (including about people)." It doesn't say whether a granted request changes the model or only the replies.

What does Google's "Results about you" remove?

Search results only. Results about you finds pages in Google Search that show your home address, phone number, or email address, and lets you select Request to remove on each one. Open it from your profile picture in the Google app, choose Get started, and enter the details you want Google to look for (Google).

Google says that "whether your request is approved or denied, the info still exists on the source web page." It won't remove results from government, educational, news, or business sites. The help page doesn't mention Gemini, so treat it as cleaning up Search, and file the Gemini request separately.

How do you ask Meta AI to stop showing your personal information?

With Meta's form for personal information from third parties. Meta defines that information as what is "publicly available on the internet" plus licensed data, and says up front that it doesn't link it to any Meta account. "As a result, we don't provide the ability to access, download, correct or delete personal information from third parties," the form reads (Meta). What it takes is a concern about your information appearing in a Meta AI answer.

The steps, as the form stood on November 5, 2026:

1. Open the form and tick the box that says you have a concern about your personal information from third parties in a response from an AI at Meta model.
2. Pick your country of residence and, in the US, your state.
3. Enter your first name, last name, and email address.
4. Paste the prompts that brought up your information under Relevant Prompts.
5. Upload a screenshot of the response, under 4 MB, as JPG, PNG, GIF, TIFF, HEIF, or WebP.
6. Add context, and click Submit.

Meta says "we don't automatically fulfill requests sent using this form" and reviews them "consistent with your local laws." Its explainer on generative AI adds that deleting a post a model learned from leaves the model unchanged "in that moment," though the post isn't used in later training. Your own Facebook and Instagram posts are a separate matter, covered in how to turn off Meta AI.

How do you remove your information from Claude?

Write to Anthropic. Its article on training-data requests, dated July 8, 2026, asks for your name, your email address, "screenshots or details of the response that shows your personal data, the prompt used, or the URLs referenced." Anthropic's non-user privacy policy names privacy@anthropic.com as the address and promises a GDPR answer within one calendar month, extendable by two more.

Anthropic is the most candid of the four about what it can find. It says it doesn't link training data to specific people, so "there are limitations on our ability to identify information from our training data that relates to a specific individual." What it describes instead is a way to lower the odds Claude repeats you: "techniques that can reduce the likelihood of data from our training data being reproduced by Claude." On corrections, the policy says it will make "a reasonable effort," but "it may not always be possible" given "the technical complexity of large language models." It also says it keeps training data as long as "the models trained on it remain in service or supported."

Does the right to be forgotten apply to AI?

In the EU and UK, the right exists, with exceptions the companies use. GDPR Article 17 lists six grounds for erasure, and Article 16 gives you "the rectification of inaccurate personal data." Article 17(3) then sets out five exceptions, starting with processing needed "for exercising the right of freedom of expression and information" and including legal obligations, public health, research and archiving, and legal claims. That first exception is the public-interest test OpenAI applies when it turns a request down, and Anthropic's training-data article gives "a public interest in the information staying available" as one reason it may decline. Whether erasure can reach a trained model is a separate question, covered in our explainer on machine unlearning.

In the US, the closest thing is a state privacy law, and California's has a gap for this exact case. The Attorney General's CCPA page, updated August 28, 2026, says the right to delete covers personal information a business "collected from" you, and that "personal information does not include publicly available information." Its example of a reason a business may keep your data is "your address, which is often in public real estate/property records." Much of what a model learned from the open web may fall under that exemption. The right to correct "inaccurate information that they have about you" is the stronger CCPA tool here, and businesses must answer within 45 calendar days, or 90 if they tell you. This isn't legal advice.

Does a removal request take your data out of the model?

Not that any of them promise: as of November 5, 2026, OpenAI describes keeping your information out of ChatGPT's responses, Meta won't correct or delete third-party data at all, Anthropic describes lowering the odds that Claude repeats it, and Google doesn't say. Our explainer on machine unlearning covers why a trained model stays as it is.

How Morse keeps your own chats from becoming data about you

The forms above all work after the fact, on information a company already holds. Morse can't change what GPT, Claude, or Gemini learned from the web, so use the forms for that. You can control the next round. Every detail you type about yourself into a chatbot that keeps and trains on chats becomes one more record about you on its servers.

Morse is built so your chats don't become that record. Morse doesn't train AI on your content, sell your data, or show ads. Your chat history is stored in your browser, encrypted, and our servers don't keep messages or log what you write, so there's no account full of your conversations to file a request against. Because we don't store chats, we usually have very little to give authorities. Clearing your browser, losing the device, or deleting your account deletes the chats, and Morse can't recover them, so save anything you need.

Pick a mode for each chat by how far your details should travel:

• Private, for questions full of your own details. It's the default for a new chat. Open-weight models like DeepSeek, Kimi, and GLM, plus Grok, answer under zero data retention, because Morse has zero data retention agreements with its inference, search, and other providers.
• E2EE, for anything about you that no one else should read. Your browser checks the enclave's hardware attestation before anything leaves your device, then encrypts the message so only that enclave can read it, and our servers relay ciphertext they can't read. The enclaves are run by third-party providers, not by Morse, and the guarantee rests on the enclave hardware and attestation working as designed.
• Anonymous, to ask GPT, Claude, or Gemini without the provider knowing it's you. Our proxy removes your account, email, and IP address and sends the message under Morse's own account. The provider sees the message, may keep it for a limited time to check for abuse under its own policy, and can still read any name you type, so leave out details you'd rather it never had.

Our privacy policy has the details. Try a chat from the Morse home page without an account, or start free with no card. End-to-end encrypted chats come with Pro, at $19 a month.