How to Remove Your Personal Data from ChatGPT (2026 Guide)

There are two separate problems people mean when they say "remove my data from ChatGPT." One is stopping OpenAI from using your conversations to train future models. The other, and the one this guide focuses on, is getting OpenAI to stop a model from surfacing your personal information in its answers to other people. These require different requests, and most guides only cover the first one.

Key Takeaways

• To stop future training on your conversations, disable "Improve the model for everyone" under Settings → Data Controls. This is forward-looking only and does not touch what's already been trained.
• To request that ChatGPT stop generating specific personal information about you in its responses, use the dedicated "Remove personal data from ChatGPT responses" request at OpenAI's Privacy Portal — a separate mechanism from the training toggle.
• The personal-data-removal request requires email verification, specific examples of the problematic output, the URLs or context where the information originated, and government-issued identification.
• OpenAI applies a balancing test against public interest and freedom of expression — public figures or people involved in matters of public record (criminal convictions, public disputes) may have removal requests denied if the output is deemed part of legitimate public debate.
• Neither request removes data already used to train existing model weights — that influence on the model cannot be surgically undone; only future outputs and future training are addressed.

The Two Different "Remove My Data" Requests

1. Stop future training on your conversations (prevention, not removal)

Go to Settings → Data Controls and toggle off "Improve the model for everyone." This prevents your future conversations from being used to train subsequent model versions. It does not delete anything already used in past training, and it doesn't affect what the model already "knows" or might say about you if you're a public figure whose information came from other sources.

2. Stop the model from generating specific personal information about you (a right-to-be-forgotten-style request)

This is the request most people actually want and don't know exists. If ChatGPT generates inaccurate or unwanted personal information about you in response to someone else's prompt — an address, a claim about your background, outdated or false information — you can submit a formal request through OpenAI's Privacy Portal asking that the model be prevented from producing that specific output going forward.

How to Submit a "Remove Personal Data from ChatGPT Responses" Request

Step 1: Go to the OpenAI Privacy Portal

Navigate to privacy.openai.com.

Step 2: Select the personal-data-removal request type

Choose the option for removing personal data from ChatGPT's responses (distinct from the general "do not train on my content" request, which only affects future training).

Step 3: Verify your email

OpenAI requires email verification before processing the request.

Step 4: Provide specific examples

You'll need to document the specific prompts that generate the problematic output, and describe or link to the source of the information if you can identify where it originated.

Step 5: Submit government-issued identification

OpenAI requires ID verification to confirm you are the person the request concerns, reducing the risk of the mechanism being abused to suppress information about someone else.

Step 6: Understand the review standard

OpenAI evaluates these requests against a balancing test that weighs your privacy interest against public interest and freedom of expression. If you're a public figure, or the information relates to a matter of public record (a criminal conviction, a public legal dispute, published news coverage), your request may be denied on the grounds that the output reflects legitimate public information rather than a private-data violation.

What This Does and Doesn't Cover

Covered:

• Future outputs where the model would otherwise generate the specific flagged personal information about you
• Your own stored conversation history and account data (deletable separately via Settings → Data Controls → "Delete all chats")

Not covered:

• Data already baked into the weights of previously trained models — there is no technical mechanism to "un-train" a model on a specific fact
• Information about you generated through general knowledge the model has about public figures or publicly documented events
• Third-party apps built on the OpenAI API — their data handling is governed by that developer's own privacy practices, not your ChatGPT account settings
• Data brokers and people-search sites that are a separate, and often larger, source of the same information appearing elsewhere online

Why This Doesn't Solve the Underlying Exposure

Even a successful ChatGPT removal request only affects one product. The same personal information the model may have picked up during training frequently originates from data broker profiles: Whitepages, Spokeo, BeenVerified, and similar sites that are scraped and licensed as training-data inputs across the AI industry. Removing your data from ChatGPT's outputs doesn't touch the broker profiles feeding that exposure in the first place, and it doesn't prevent the same information from surfacing in a different AI product entirely.

Frequently Asked Questions

How do I get ChatGPT to stop training on my conversations?

Go to Settings → Data Controls and disable "Improve the model for everyone." This is forward-looking only; it does not affect conversations already used in past training or your account's stored history unless you separately delete it.

How is removing personal data from ChatGPT's responses different from opting out of training?

The training opt-out stops your own conversations from becoming future training data. The personal-data-removal request is about stopping the model from generating specific information about you in response to someone else's questions — a separate mechanism submitted through privacy.openai.com that requires documented examples and ID verification.

What do I need to submit a personal-data-removal request to OpenAI?

A verified email address, specific examples of the prompts and outputs containing the information, context on where the information originated if known, and a government-issued ID.

Can OpenAI deny my removal request?

Yes. OpenAI weighs these requests against public interest and freedom of expression. Public figures or people connected to matters of public record may be denied if the output is considered legitimate public information rather than a privacy violation.

Does removing data from ChatGPT stop other AI products from saying the same thing about me?

No. Each AI company (Google Gemini, Anthropic Claude, xAI Grok, Microsoft Copilot) operates a separate model and, where offered, a separate removal request process. Removing personal information from one product's outputs has no effect on the others.

Related Guides

• Opt Out of AI Training Data: ChatGPT, Claude, Gemini, Grok
• How to Remove Your Personal Data from Google Gemini
• How AI Companies Scrape and Sell Your Data
• Complete Data Broker Opt-Out Guide