AI Reputation · Search Result Repair
You asked ChatGPT about yourself, or a client, a recruiter, or a prospect did, and the answer that came back belongs to someone else. Someone with your name who has a criminal record. A lawsuit. A bankruptcy. A news story you had nothing to do with. The AI did not hedge or flag any uncertainty. It presented another person’s history as yours, in fluent, confident prose, to whoever asked.
This is one of the most jarring reputation problems a person can encounter, precisely because you did nothing to cause it. And it is becoming common, because more people now ask an AI assistant about a name than run a traditional background search on it.
Know this first.
- AI assembles the most statistically plausible answer for a name, not a verified record.
- A shared name, a thin footprint, and inconsistent signals make the mix-up almost certain.
- The same ambiguity feeds every AI system, so fixing one product does not fix the record.
- The durable fix is disambiguation: consistent, authoritative material that separates you from the other person.
What is probably happening
Search strategists call it entity confusion.
AI systems do not look you up in a verified registry. There is no master file that says which John Marsh is which. When a model is asked about a name, it draws on patterns from its training data and, in many cases, live web results, then assembles the most statistically plausible answer. If two people share a name, and one of them generated more coverage, more indexed pages, more discussion, the model’s answer gets pulled toward that person. Court records, news stories, and controversy generate a lot of coverage. Quiet professional lives generate very little.
Search strategists call this entity confusion. The model has failed to maintain a boundary between two distinct people, and it merges their attributes: your job title with their arrest, your city with their lawsuit, your photo era with their timeline. The blend can be seamless, which is what makes it dangerous. A reader has no way to see the seam.
Three conditions make confusion much more likely. A common or shared name, obviously. A thin online footprint, because if there is little authoritative material about you, the model fills the vacuum with whoever else it can find. And inconsistent identity signals, meaning your name appears in different forms across different sites, with no stable set of facts, employer, location, field, connecting them. Any one of these invites the mix-up. All three together practically guarantee it.
Why this shows up everywhere, not just ChatGPT
The confusion rarely lives in one system. ChatGPT, Gemini, Perplexity, Claude, and Google’s AI features all lean on overlapping source material: indexed web pages, news archives, public records aggregators, and the profiles that rank for your name. If the public record about your name is ambiguous, every system reading it inherits the ambiguity. That is why the fix is not complaining to one AI company. It is repairing the record they all read. For the broader problem, see our guide to what to do when AI is saying false things about you.
It is also why the problem compounds quietly. You will almost never be told this happened. A hiring manager asks an assistant about a candidate, sees a liability, and simply moves on. A sales prospect checks a name before a call and goes cold. The confusion costs you opportunities you never knew existed.
What not to do
Do not assume one correction fixes it. ChatGPT offers ways to report inaccurate outputs about yourself, and OpenAI has privacy request channels. Use them. But a correction in one product does not repair the underlying record, and the same confusion will keep resurfacing in other systems, and sometimes in the same system, asked a slightly different way.
Do not publish a public rebuttal that repeats the false association. Writing “I am not the David Chen who was convicted of fraud” on your website ties your name to that phrase in exactly the way you do not want. You would be feeding the association you are trying to break.
Do not scrub yourself off the internet. Deleting your LinkedIn and locking down every profile feels protective, but it removes the only competing signal. The less verified material exists about the real you, the more completely the other person’s record fills the space. Invisibility is what caused this.
What can sometimes be corrected
Some pieces of this have direct remedies. AI providers maintain feedback and privacy request processes for factually false statements about real people, and these are worth filing, especially for the most damaging specific claims. People-search and data broker sites that have merged your records with someone else’s often have opt-out or correction processes. If a specific web page falsely identifies you, the site owner can sometimes be persuaded to correct it, and content that is defamatory may have removal paths worth discussing with an attorney, though that is a legal question, not something we advise on.
These corrections matter, but understand their limits. They treat instances, not the cause. The cause is that the internet does not clearly distinguish you from the other person, and no takedown fixes that.
What usually needs rebuilding
The durable fix is disambiguation: building a body of consistent, authoritative material that makes it easy for both search engines and AI systems to tell the two of you apart. In practice that looks like a personal website that states plainly who you are, what you do, and where, with your name used in one consistent form. Profiles on the platforms that carry weight in your field, all aligned on the same facts. Structured signals, the same photo, the same title, the same biographical anchors, repeated across every property. And ideally, credible third-party material, interviews, professional listings, publications, bylines, that reinforces the distinct identity from outside your own properties.
Done consistently, this gives the models the boundary they were missing. Your cluster of signals becomes coherent and self-confirming, and the other person’s record stops bleeding into it. This is slower than a takedown, and it is the part people underestimate, but it is what actually holds. For private individuals who have never needed a public presence before, our reputation repair for private individuals page covers how this is done quietly and proportionately.
See exactly what AI says for your name
We document what the major AI systems and search results show for your name, what can realistically be corrected, and what needs to be rebuilt so the confusion cannot recur.
When to get a private assessment
Get an assessment before you act if any of these apply: the confused-with person has a serious criminal or legal history; the confusion is showing up in more than one AI system or in Google itself; you are actively job hunting, fundraising, selling, or dating, meaning people are checking your name right now; or your own footprint is so thin that you have nothing established to build on. In those situations the sequencing matters, what to correct first, what to build first, what to leave untouched, and a wrong early move, like the public rebuttal, can set you back.
Frequently asked questions
Can I make ChatGPT stop talking about me entirely?
You can submit privacy requests to AI providers, and in some jurisdictions you have formal rights to object to processing of your personal data. Some requests result in a model declining to answer about a name. But this is not guaranteed, it does not apply to every system, and for professionals it can backfire, because silence about your name leaves whatever ranks in search as the only answer.
Is this defamation? Can I sue an AI company?
That is an attorney question, and the law here is genuinely unsettled. What we can say from the search side: legal processes are slow and public, and even a successful outcome against one company does not correct the shared source material that other systems read. Most people run the practical repair in parallel with, or instead of, any legal route.
How long does it take for AI systems to stop confusing me with the other person?
Systems that read the live web can improve within weeks to a few months of the record being repaired, because they reflect current search results. Corrections to training data move slower and depend on each provider’s update cycles. Honest answer: expect gradual improvement as your signals strengthen, not a single flip.
I have a very common name. Is this even solvable?
Yes, though the goal shifts. With a very common name you are not trying to own the name outright; you are trying to make your identity unmistakable in your context, your field, your city, your qualifier. Models handle “Sarah Miller the Denver structural engineer” far better than an unanchored “Sarah Miller,” but only if that anchoring exists consistently online.
Get clarity before you respond
Being mistaken for someone else is not a small glitch. It is the public record of your life being overwritten by a stranger’s, and the instinctive responses, public denials, deleting profiles, one-off complaints, tend to make it worse or leave it unfixed. Before you do anything, get a private assessment of exactly what the AI systems and search results are showing for your name, what can realistically be corrected, and what needs to be rebuilt so the confusion cannot recur. Reach out privately and we will tell you what your situation actually requires.
Mistaken for someone else by AI?
Send us your name. We will document what the AI systems show, what can realistically be corrected, and what to build so the confusion cannot recur.
Private. Confidential. No public case studies. No pressure.
