Written by the Removal Lab team.
A prospective client asks ChatGPT whether your firm is any good. A patient asks Gemini about your clinic. A recruiter asks Copilot about you by name. The answer arrives in two paragraphs, sounds confident and may be built on a single bad review from years ago. AI reputation management is the work of finding out what these assistants say, understanding where they got it and fixing it at the source. Removal Lab offers it as a free check first, because most people have no idea what the machines are saying about them until a customer mentions it.
How AI assistants decide what to say about you
Every major assistant now searches the live web for questions about a specific business or person. ChatGPT with search, Gemini, Copilot, Perplexity and Google AI Overviews all pull from pages they find, summarise them and often cite them. They favour pages that rank well, answer the question plainly and look trustworthy. That is why AI reputation management is mostly about the same things as search: what ranks for your name and what it says.
Two sources carry unusual weight. Review platforms, because they are structured and the assistants treat star ratings as facts. And any page that states something about you in a plain declarative sentence, because assistants repeat those sentences almost word for word. A review that says “this dentist overcharges and the hygienist was rude” can become “some reviewers report overcharging and rude staff” in an answer read by every prospective patient who asks.
The uncomfortable truth
Assistants do not check whether a review was fake. They check whether it ranks. If the fake review is on page one for your name, it is in the answer.
What AI reputation management involves
| Step | What we do | What it changes |
|---|---|---|
| The AI answer check | Ask ChatGPT, Gemini, Copilot, Perplexity and Google AI Overviews the questions your customers ask, in each market you serve | You see exactly what is being said and which sources are cited |
| Source mapping | Trace every claim back to the page it came from | You know which review, article or listing is the real problem |
| Source removal | Report reviews and pages that break platform rules or Google policy, with evidence | The claim disappears from the next answer when the source does |
| Source correction | Ask editors, directories and platforms to correct inaccurate facts | Wrong addresses, closed businesses, old names and misattributed events get fixed |
| Structured content | Build pages that state the facts about you plainly, with the right schema markup | Assistants have a reliable source to quote instead of a stray review |
| Platform reports | File personal data requests and inaccuracy reports with OpenAI, Google and Perplexity where the answer itself is the problem | Covers the cases where the source is the model, not a web page |
| Monitoring | Re run the questions monthly and report changes | You hear about a new problem before customers do |
Why removal still matters in the age of AI
Some agencies now pitch AI reputation management as a content exercise only. Publish enough good pages and the assistants will quote those instead. That is half true and the half that is false is expensive. A single fake review on Google carries more weight with an assistant than ten blog posts, because the assistant trusts the review platform more than it trusts you. If that review breaks Google’s rules, the right move is to report it and have it removed. That costs nothing unless it works. The content work comes after, to fill the space with something true. Our fake review removal page explains what qualifies.
The questions we ask the assistants
A good AI answer check asks the questions real customers ask, not vanity questions. For a law firm in Toronto, that means “is this firm reputable”, “what do clients say about this firm”, “who is the best family lawyer in Toronto” and “should I hire this firm”. For an individual, it means “who is this person”, “what is this person known for” and “is there anything concerning about this person”. We ask each question in each assistant and each market. We record the answer and the citations. The report shows you the pattern, which is usually two or three sources doing all the damage.
What can be fixed and what cannot
| Situation | Can it be changed? | How |
|---|---|---|
| Answer repeats a review that breaks platform rules | Usually | Review removal under the platform’s policy |
| Answer repeats a genuine negative review | Partly | Response on the platform, more recent genuine reviews, content that gives the assistant a fuller picture |
| Answer states a wrong fact from a directory or old listing | Usually | Correction at the source, outdated content tool for Google |
| Answer repeats a news story | Rarely at source | Legal routes where they exist, otherwise content that provides context |
| Answer invents something that is not on any page | Sometimes | Inaccuracy report to the AI provider, plus clear pages stating the facts |
| Answer includes personal contact details | Usually | Google personal information removal, source opt outs and AI provider personal data forms |
| Answer confuses you with someone else of the same name | Often | Clear disambiguating content with schema markup, correction requests |
We are candid about the limits. Assistants that rely on training data rather than live search can repeat an old claim for months after the source is gone. There is no form that fixes that overnight. AI reputation management is patient work and we label every timeline as a projection, not a promise.
AI reputation management on each engine
Each assistant explains its sources a little differently. Here is what each provider’s own documentation says about where answers come from and what a business or a person can do when an answer is wrong. Removal Lab works from these pages on every AI reputation management case and we link them at the end. Where a provider does not publish a correction route, we say so.
ChatGPT
OpenAI’s help page on ChatGPT search says responses that use search contain inline citations and that the Sources button at the end of a response lists every cited page. It also says ChatGPT may share disassociated search queries with third party search providers such as Bing. In plain terms, the pages that answer a web search about you are often the pages ChatGPT reads.
OpenAI’s crawler documentation describes three agents. Its search crawler decides whether a site can appear in ChatGPT search. OpenAI says sites that block it will not be shown in ChatGPT search answers. A separate crawler collects content that may be used to train its models. A third agent visits a page when a user asks a question. OpenAI says robots.txt rules may not apply to that agent because a user started the visit. OpenAI says a robots.txt change can take about 24 hours to reach its systems.
That matters for AI reputation management. Blocking OpenAI’s search crawler on your own website removes your side of the story from ChatGPT search answers, while the critical pages stay in. We usually advise the opposite: let it in and make sure your own pages state the facts.
To correct personal information, OpenAI’s article on how its models are developed says training data includes publicly available information, licensed information and information from users and trainers. It adds that this incidentally includes personal information. It says people may have rights to access, correct, restrict, delete or transfer that information. OpenAI’s privacy policy says some rights can be used through your OpenAI account and the rest by emailing dsar@openai.com. For a business, the faster AI reputation management fix is nearly always the source page ChatGPT cites.
Google AI Overviews and AI Mode
Google’s Search Central page AI features and your website says AI Overviews and AI Mode may use query fan out, which means issuing multiple related searches across subtopics and data sources. To appear as a supporting link, a page must be indexed and eligible to show in Search with a snippet. Google says there are no additional requirements and no special schema markup to add.
Google’s guide to optimising for generative AI features says these features are rooted in its core Search ranking and quality systems. It says Google Search itself does not use llms.txt files, that there is no ideal page length and that Business Profiles and Merchant Center feeds help products and services show in AI responses. So AI reputation management on Google starts with ordinary things: accurate indexed pages and a correct Business Profile.
To report a wrong answer, Google’s AI Overviews help page says to click the thumbs down icon at the bottom of the overview, select Report a problem, choose a category, add details if you wish and click Submit. The AI Mode help page describes the same thumbs down, category and Submit steps. Google says the feedback includes your most recent search and its results. Feedback is worth sending. In AI reputation management it works best alongside a fix to the page the answer relied on.
For phone numbers, home addresses and email addresses in Search results, Google’s personal information policy and the Results about you tool are the official routes. Site owners can also limit what Google shows from their own pages with the nosnippet and noindex controls described on the AI features page.
Gemini
Google’s Gemini Apps privacy hub says some Gemini responses are grounded on Search results and warns that Gemini can make mistakes. The same pages that shape Google Search therefore shape many Gemini answers about you.
Below a Gemini response, select More and then Report legal issue. Google also lets you create a request in its Help Center to exercise your data rights and to request removal of content under its policies or applicable laws. For ordinary errors, Gemini has a send feedback option. In AI reputation management we pair any Gemini report with a fix to the page that fed the answer, because a report on its own does not change the source.
Perplexity
Perplexity’s help centre says it searches the internet in real time and that each answer includes numbered citations linking to the original sources. Its crawler documentation says its search crawler is designed to surface and link websites in Perplexity search results and is not used to crawl content for AI foundation models. A second agent visits a page when a user asks a question. Perplexity says that agent generally ignores robots.txt because a user requested the visit.
Perplexity advises sites to allow its search crawler in robots.txt to appear in results. The Perplexity pages we read do not describe a form for correcting an answer about a business. Its privacy policy lets people ask for access to, correction of or deletion of personal data through account settings or a data request form. In practice, AI reputation management on Perplexity means reading the numbered citations and fixing or removing the pages behind them.
Microsoft Copilot
Bing’s Webmaster Tools blog says its AI Performance report covers Microsoft Copilot, AI generated summaries in Bing and select partner integrations. It shows how often your content is cited as a source in AI answers, which of your pages are cited and a sample of the grounding queries used. Bing says it respects content owner preferences expressed through robots.txt. For AI reputation management, this report shows which of your pages Copilot already cites.
The same post recommends IndexNow to tell participating search engines when content is added, updated or removed, which helps keep AI answers current. It also recommends Bing Places for Business so a local address, hours and contact details stay correct. A Bing product manager writing on the Microsoft Advertising blog advises clear headings, direct question and answer pairs, tables and schema markup. The same article warns against hiding key answers in tabs, PDFs or images. The Microsoft pages we read do not describe a correction form for Copilot answers, so AI reputation management on Copilot runs through your sources and Bing.
Where each AI engine gets its answers
A quick reference for AI reputation management, built only from the provider pages linked in this section.
| Engine | Where answers come from | What you can do |
|---|---|---|
| ChatGPT | Live web search with inline citations and a Sources list. OpenAI says it may share disassociated search queries with third party search providers such as Bing. Model training uses public, licensed and user provided information. | Fix or remove the pages it cites. Keep OpenAI’s search crawler allowed on your own site. Use OpenAI’s privacy rights routes for personal data. |
| Google AI Overviews and AI Mode | Google’s core Search ranking and quality systems, with query fan out across related searches. A supporting page must be indexed and eligible to show with a snippet. | Thumbs down, then Report a problem. Personal information removal and Results about you for contact details. Keep your Business Profile accurate. |
| Gemini | Google says some Gemini responses are grounded on Search results. | More, then Report legal issue. A Help Center request for removal under Google’s policies or the law. Send feedback for general errors. |
| Perplexity | Real time web search. Each answer carries numbered citations to the original sources. | Check the citations and fix the sources. Allow Perplexity’s search crawler. Privacy requests through account settings or its data request form. |
| Microsoft Copilot | Bing, which shows your content as a source or reference in AI answers and respects robots.txt. | Bing Webmaster Tools AI Performance report to see cited pages, IndexNow after you fix a page, Bing Places for local details. |
What nobody can do
No agency, Removal Lab included, can edit an AI model’s answer directly or pay an assistant to say something different. The providers describe feedback, privacy and legal routes. Everything else in AI reputation management is about the sources: removing the ones that break the rules, correcting the ones that are wrong and publishing pages that are true. For people in the public eye, our executive reputation management page covers the extra privacy steps.
Reporting to the AI providers
OpenAI, Google and Perplexity each publish routes for personal data requests. OpenAI’s privacy policy lists rights to correct or delete personal information. Google accepts feedback on AI Overviews and AI Mode through Report a problem, Gemini has a Report legal issue option and Google Search has its personal information removal process. The Microsoft pages we read do not describe a correction form for Copilot, so for Copilot we work through the sources and Bing. These routes are useful for personal data and for plainly false claims about an individual. They are not a route for businesses that dislike a summary of genuine reviews. We file them where they fit and we tell you when they do not.
Building the pages the assistants prefer
Assistants reward pages that answer questions plainly. A page titled with the question, a short direct answer in the first paragraph, the detail below, the facts in a table and the right schema markup for an organisation, a person or a service. Your own website is the best place for this. Properly built, it becomes the page the assistant quotes when someone asks who you are and what you do. This is the constructive side of AI reputation management and it is the same work that earns Google rankings, because the assistants and Google reward the same qualities.
What does not work: pages stuffed with your name, press releases nobody reads, dozens of thin profiles and anything that reads as written for machines. The assistants are good at spotting that and so is Google.
An illustration of how AI reputation management works
This is a made up example to show the steps. It is not a client case.
Imagine a dental practice in New York asking why new patient enquiries have slowed. The Google rating was 4.6. The website was fine. We ran the AI answer check. Gemini and Google AI Overviews both answered “is this practice good” with a balanced summary. ChatGPT answered with a paragraph that led on “several reviewers mention being charged for procedures they did not agree to”, citing one Yelp review from a person the practice had no record of and one Google review from a former employee.
| Finding | Route | What changed |
|---|---|---|
| Yelp review from a non patient | Reported under Yelp’s first hand experience guideline with the practice’s confirmation that no record existed | Removed by Yelp |
| Google review from a former employee | Reported under conflict of interest with the pattern of timing and wording | Removed by Google after appeal |
| No page on the website stating fees policy plainly | Built a short, clear page on how fees are agreed before treatment, with FAQ schema | Gave assistants a direct source for the exact question patients ask |
| ChatGPT answer unchanged for three weeks after removals | Monitored; answer updated once the live search no longer found the reviews | Answer now leads on the rating and the fees policy page |
In a real case nothing like that is promised. Yelp and Google make the decisions and the assistant updates on its own schedule. The point is the method, which is the whole of AI reputation management: find the sources, fix the sources, give the machines something true to quote.
AI reputation management and ordinary SEO
People ask whether this replaces search optimisation. It does not. It sits on top of it. The assistants read the same web Google reads and prefer the same things: pages that rank, answer plainly, load fast and carry accurate structured data. A business that has done the ordinary work well is already most of the way there. What AI reputation management adds is the check itself, the source mapping and the specific fixes for the way assistants summarise. If you have never asked an assistant about yourself, do it now in a private window. Read the answer as a customer would. Then send it to us.
Who needs AI reputation management
| Who | Why it matters now |
|---|---|
| Law firms, clinics and professional practices | Clients ask assistants for recommendations and the answer quotes whichever review ranks first |
| Restaurants, hotels and tour operators | Travel questions are among the most common uses of assistants and TripAdvisor is a favourite source |
| Executives and founders | Due diligence now starts with an assistant, not a search engine |
| Anyone with a common name | Assistants merge people with the same name and attribute the wrong events |
| Anyone who has had a bad review removed | The assistant may still repeat it from cached or training data until the answer is checked and the gap is filled |
What AI reputation management costs with Removal Lab
The AI answer check is free. It takes a few days and you receive a plain report showing the answers, the sources and what we recommend for each. Removal of sources that break platform rules is pay on results: no upfront payment on qualifying cases and nothing owed if the platform says no. Correction requests are included in any removal engagement. Structured content and monitoring are monthly plans quoted after the check, with the scope and the reporting written down. We never post fake reviews, fake articles or fake profiles, because they break the rules, break the FTC rule on consumer reviews in the United States and would put your own profile at risk.
Where we work
Removal Lab serves clients in Canada, the United States, the United Kingdom, Australia, the United Arab Emirates and Switzerland from offices in Toronto and New York. Assistants give different answers in different countries, so we run the check in each market you serve.
How Removal Lab works
- Send the link. WhatsApp, email or the two field form. Two minutes, no long questionnaire.
- Free policy check. A specialist reads it against the platform rules and tells you honestly whether it qualifies.
- We build and file the case. The right policy, the right channel, the evidence attached. We handle the appeal too.
- Pay only when it is gone. No upfront payment on qualifying cases. If the platform says no, you owe nothing.
What we will never do
- Promise a removal. The platform decides, not us.
- Post fake reviews or pay anyone to delete a review.
- Pressure, threaten or try to identify a reviewer.
- Charge an upfront fee on a case that does not qualify.
Related reading
- Online reputation management services
- Google search result removal
- Fake review removal
- What we cannot remove
- Suppress negative search results
- Remove personal information from Google
Frequently asked questions
What is AI reputation management?
It is the work of finding out what AI assistants such as ChatGPT, Gemini, Copilot and Google AI Overviews say when someone asks about you or your business, understanding which web sources they are drawing on and fixing those sources.
Can you remove something from ChatGPT?
Not directly. AI assistants build answers from web sources and from training data. The practical route is to remove or correct the source pages, which changes what the assistant finds the next time it looks. OpenAI’s privacy policy also lists rights to correct or delete personal information. Google’s AI features have a Report a problem option.
Why does an AI assistant repeat a bad review?
Because the review ranks well for your name and the assistant treats it as a source. Review removal under the platform’s rules takes the source away. Good content that outranks it changes what the assistant sees first.
Does AI reputation management replace SEO?
No. AI assistants cite the pages that rank well and read clearly. Honest, well structured pages about your business are what both Google and the assistants prefer.
How long does it take to change an AI answer?
Assistants that search the live web change within days of the sources changing. Answers drawn from training data change only when the model is updated, which can take months. We label every timeline as a projection, not a promise.
What does Removal Lab charge?
The AI answer check is free. Removal of sources that break platform rules is pay on results with no upfront payment on qualifying cases. Ongoing content and monitoring are monthly plans quoted after the check.
How do I report a wrong Google AI Overview about my business?
Click the thumbs down icon at the bottom of the AI Overview, select Report a problem, choose a category, add details and click Submit. Google says the feedback includes your latest search and its results, so fix the source page as well, because AI reputation management works best at the source.
Does an llms.txt file change what AI says about me?
Not on Google. Google’s guide to generative AI features says Google Search itself does not use llms.txt files and that no special schema markup is needed. Accurate indexed pages, a correct Business Profile and fixing the pages assistants cite matter far more.
Sources we work from: Google policy on removing personal information from search, OpenAI privacy request portal, Google Maps prohibited and restricted content, FTC rule on consumer reviews and testimonials, OpenAI: ChatGPT search, OpenAI crawlers, OpenAI: how its models are developed, OpenAI privacy policy, Google Search Central: AI features and your website, Google’s guide to optimising for generative AI features, Google: AI Overviews help, Google: AI Mode help, Google: Results about you, Gemini Apps privacy hub, Perplexity: how Perplexity works, Perplexity crawlers, Perplexity privacy policy, Bing Webmaster blog: AI Performance, Microsoft Advertising blog: inclusion in AI search answers and IndexNow. Removal Lab does not work for any AI provider or platform. Every decision is made by the provider or platform concerned.