ChatGPT Says Something Wrong About You
OpenAI now has a formal process for asking that certain personal information stop appearing in ChatGPT responses. Here’s what qualifies, what evidence matters and what the request does not erase.
A few years ago, if Google had the wrong information about you, at least you could usually trace the problem back to a webpage. AI makes that messier. A person can ask ChatGPT who you are and get a neat, confident paragraph containing an old job, a mixed-up namesake, a stale allegation or a fact that simply isn’t true. The important change is that you no longer have to treat that answer as untouchable. OpenAI now provides a specific privacy process for asking that personal information about you stop appearing in ChatGPT responses.
Yes, there is now a specific request for this
OpenAI’s Privacy Portal includes an option called “Remove personal data from ChatGPT responses.” It is separate from deleting your account, deleting a conversation or turning off model training.
Route
You do not need a ChatGPT account
OpenAI says a person can ask it to prevent information about them from appearing in ChatGPT responses even if they do not have an account. A legally authorized representative can also make a request, although OpenAI may require proof of that authority.
Find the bad output
Capture the response, the prompt that produced it and any cited URLs connected to the information.
Submit a privacy request
Use OpenAI’s Privacy Portal and choose “Remove personal data from ChatGPT responses.”
Make the evidence easy to judge
Explain exactly what is wrong, outdated, excessive or irrelevant and provide reliable supporting evidence.
First make sure you’re fixing the right problem
Four completely different issues often get described as “ChatGPT has my information.” Only one is the reputation-removal problem this report is about.
| Problem | What is happening | Correct first move |
|---|---|---|
| ChatGPT tells other users incorrect personal information about you | The problem is information appearing in generated responses. | Privacy Portal removal request |
| A website cited by ChatGPT contains the wrong information | The underlying public source itself is inaccurate. | Correct the source + consider the OpenAI request |
| You typed private information into your own ChatGPT account | This is primarily an account-data or conversation issue. | Use the appropriate ChatGPT data controls |
| You want a Google page removed | The material exists on an external site or search engine. | The OpenAI request does not remove it from Google |
Removing it from ChatGPT does not erase the internet.
OpenAI explicitly says that removing personal data from ChatGPT responses does not remove the information from websites or search engines. If the original article, database, court page, profile or forum post is wrong, that source remains a separate reputation problem.
What kinds of information can you challenge?
OpenAI’s current guidance centers on personal information that is inaccurate, excessive, irrelevant or no longer appropriate.
| Example | Why it may matter | Best evidence |
|---|---|---|
| Wrong employer, profession or biography | Clearly inaccurate identity information can misrepresent who you are. | Current employer page, professional profile or other authoritative biography. |
| Information belonging to somebody with the same name | This is an identity accuracy problem rather than merely unwanted information. | Pages clearly distinguishing the two people. |
| An allegation presented without its later outcome | An answer can become misleading when important context is omitted. | Court disposition, official decision or authoritative later reporting. |
| Very old personal information | Age and ongoing relevance are factors OpenAI says it considers. | Dates plus evidence showing the circumstances materially changed. |
| Sensitive private information | The nature and sensitivity of information can weigh in the privacy analysis. | Identify the precise information and explain the privacy impact. |
| True information tied directly to an important public role | Public interest and professional relevance may weigh against removal. | Context still matters, but removal is not automatic merely because the information is unwanted. |
“I don’t like this answer” is not the strongest argument. The better request identifies the exact personal information, explains why it is inaccurate, misleading, excessive, irrelevant or outdated, and backs that explanation with something an outside reviewer can verify.
The factors OpenAI says it reviews
This is useful because the process is not a black box. OpenAI has published the considerations its privacy team may use when evaluating a removal request.
The strongest way to submit the request
Treat it less like a complaint and more like a miniature evidence file. Make the reviewer understand the problem in two minutes.
Reproduce the answer cleanly
Before filing anything, confirm that the problem can actually be reproduced. Use a fresh or unpersonalized ChatGPT conversation where possible so information remembered from your own previous conversations does not get confused with a broader reputation problem.
Identify the exact statement
Do not submit ten paragraphs and say “this is wrong.” Point to the specific name, occupation, allegation, address, relationship, date, URL or other personal information you want reviewed.
Save every cited source
If ChatGPT searched the web, capture the URLs it relied on. A bad answer coming from a bad webpage often requires two repairs: the underlying webpage and the ChatGPT response.
Build evidence around the correction
OpenAI says it cannot independently investigate disputed facts. If you say something is false, provide reliable material showing why.
Explain the reason for removal
Use plain English. Tell the reviewer whether the information is inaccurate, misleading, excessive, irrelevant, outdated or particularly sensitive—and explain the real-world impact.
Weaker: “This is ruining my reputation. Delete everything about me.”
Use the dedicated Privacy Portal route
Go to OpenAI’s Privacy Portal, choose Make a Privacy Request, then select Remove personal data from ChatGPT responses.
Be prepared to verify that you’re the person involved
OpenAI says a government-issued ID or alternative proof may be required so it can verify that the information really relates to you. Authorized representatives may also be asked to establish their authority.
What a removal request does—and does not do
| Action | What it affects | What it does not necessarily affect |
|---|---|---|
| Personal-data removal request | OpenAI can review whether specified information about you should stop appearing in ChatGPT responses. | The external website where the information originated. |
| Correct the source website | Fixes the information where it originates and may improve what web-connected systems encounter later. | Does not guarantee every AI system instantly updates. |
| Delete a ChatGPT conversation | Removes that conversation from your account immediately and schedules deletion from OpenAI systems under its retention rules. | Does not function as a public reputation takedown request. |
| Turn off model improvement | Stops eligible new personal-account conversations from being used to improve OpenAI models. | Does not tell ChatGPT to stop mentioning public information about you. |
| Delete your ChatGPT account | Deletes the account and associated eligible account data under OpenAI’s retention process. | Does not remove third-party public information about you from the internet. |
What makes a request harder?
A request can be legitimate and still be denied. OpenAI’s guidance makes clear that privacy rights such as erasure or objection are not absolute.
Current public-role information
If the information directly concerns a person’s current public or professional role, public-interest considerations may carry more weight.
A genuine factual dispute
If two sides simply disagree about what happened, a reviewer needs evidence. OpenAI says it cannot independently investigate contested facts.
Current official records
Material from courts, regulators or public authorities may carry continuing public-interest value, although official status alone does not make it permanently immune from review.
Requesting removal of everything
The process is built around particular personal information and context. A precise request is much easier to assess than a blanket demand to make a person disappear.
Do not confuse privacy removal with rewriting history. If information is accurate, current and genuinely important to understanding someone’s public or professional conduct, OpenAI may decide that public-interest considerations outweigh the removal request.
A two-track reputation repair is usually stronger
When ChatGPT cites a webpage containing incorrect information, repairing only the AI answer leaves the original problem alive.
Fix the AI surface
Document the incorrect ChatGPT answer and use OpenAI’s personal-data removal process when the facts fit the privacy criteria.
Fix the source
Ask the publisher, database, employer, directory or other source to correct inaccurate information. Search engines and other AI systems can still encounter that source even if OpenAI approves your request.
OpenAI: Right to be Forgotten & Personal Data Removal OpenAI Privacy Portal OpenAI Privacy Policy How ChatGPT & OpenAI Models Are Developed OpenAI Data Controls FAQ
Is the OpenAI privacy request the right move?
This does not predict whether OpenAI will approve a request. It checks whether you have the right type of problem and whether your evidence is ready enough to submit.
