What Does ChatGPT Say About You? | The Discoverability Company

What Does ChatGPT Say About You?

Find out what AI chatbots like ChatGPT are saying about you or your business, and learn how to influence the results.

Drew Chapin
By · Founder, The Discoverability Company
Published · Updated

Have you asked ChatGPT about yourself yet? If not, you should. Millions of people are using ChatGPT to research businesses, evaluate professionals, and make purchasing decisions. What it says about you, or whether it says anything at all, is shaping perceptions right now.

Open ChatGPT and type your name or your business name. Ask it to describe you. Ask it what you are known for. Ask it to compare you to your competitors. The answers may surprise you, and understanding why ChatGPT responds the way it does is the first step toward influencing those responses.

Where ChatGPT Gets Its Information

ChatGPT is trained on a massive dataset of text from the internet, including websites, news articles, books, forums, and social media. It has a knowledge cutoff date, which means it does not know about things that happened after its training data was collected. It synthesizes what it has learned into conversational responses, but it does not search the web in real time (unless using specific browsing features).

This is fundamentally different from how Perplexity works, which searches the web live for every query. With ChatGPT, the information is baked into the model. If positive, authoritative content about you existed on the web when the training data was collected, ChatGPT likely knows about it. If it did not, ChatGPT may have limited or inaccurate information, or it may not know about you at all.

Why ChatGPT Might Get Things Wrong

ChatGPT can confidently present information that is incomplete, outdated, or outright incorrect. This is not a bug. It is a characteristic of how large language models work. They generate the most statistically likely response based on patterns in their training data, which means they can produce plausible-sounding statements that are factually wrong.

For businesses, this can be a real problem. ChatGPT might associate you with the wrong industry, attribute achievements to someone else, or miss your most important differentiators entirely. It might surface negative information that has since been resolved or present outdated details about your services.

What You Can Do About It

You cannot directly edit what ChatGPT says. But you can influence it by shaping the information landscape that future training data will draw from. This is where AI optimization intersects with traditional reputation management and SEO.

Start by building a strong, authoritative web presence. Your website should clearly describe who you are, what you do, and what you are known for. Use specific, factual language rather than vague marketing copy. The more clear, consistent information that exists about you across reputable sources, the more likely AI models are to get it right.

Third-party coverage is critical. Press articles, industry publications, Wikipedia mentions, and authoritative directory listings all contribute to the dataset that AI models train on. One mention in a respected outlet carries more weight than dozens of self-published blog posts.

I wrote about this concept in detail on HackerNoon: if your products are not AI searchable, you are already losing. The businesses that invest in building citable, authoritative content now are the ones that AI systems will represent accurately in the future.

ChatGPT vs. Other AI Models

ChatGPT is not the only AI system people are using to research businesses. Claude, built by Anthropic, handles reputation queries differently, with more caution and more explicit statements about uncertainty. Perplexity searches the web live and cites its sources directly. Google AI Overviews pull from Google's own index.

Each platform has its own tendencies, and what one says about you may differ from what another says. A comprehensive AI visibility strategy accounts for all of them, not just the most popular one.

Monitoring Your AI Presence

Make it a habit to periodically check what AI systems say about you. Ask ChatGPT, Claude, and Perplexity about your business. Note what they get right, what they get wrong, and what they leave out. This gives you a roadmap for where to invest your content and PR efforts.

Pay attention to how AI systems answer comparison queries too. "Who are the best [your industry] in [your city]?" is a query that more and more people are asking AI rather than Google. If your competitors appear in those answers and you do not, that is a visibility gap you need to close.

Take Control of Your AI Narrative

The businesses and professionals who proactively manage their AI presence will have a significant advantage over those who discover too late that AI is saying the wrong things about them. This is still early. The playing field is not yet crowded. But it will be.

If you want to understand what AI systems are saying about you and take concrete steps to shape that narrative, our AI search optimization services are built for exactly this. We audit your AI presence across all major platforms and build a strategy to improve it. Book a consultation below to get started.

Related Resources

Research and Context Behind This Guide

Public adoption of AI tools for research and decision-making has accelerated fast. A March 2025 study from Pew Research on US public and expert AI views found that a majority of American adults have now used a generative AI tool, with awareness of ChatGPT specifically sitting above 90 percent. That's not a niche audience anymore. When someone types your name into ChatGPT, they're part of a massive and growing behavior pattern that shapes real purchasing and hiring decisions.

On the technical side, the gap between how AI systems synthesize information and how traditional search indexes it is well documented in recent information retrieval literature. Preprints catalogued at arXiv's Information Retrieval section show ongoing research into how large language models weight source authority, recency, and repetition when forming responses. The practical takeaway: consistent, factually precise coverage across multiple independent domains outperforms any single high-traffic page. Meanwhile, Google Search Central's AI features guidance confirms that structured, authoritative on-page content improves how AI-powered summaries represent a business, a principle that extends to how ChatGPT itself was trained on crawled web data.

The privacy dimension is real too. The FTC's privacy and security guidance for businesses increasingly covers AI-generated profiles and inaccurate data representations, and the International Association of Privacy Professionals has published frameworks for understanding when AI-generated descriptions of individuals may trigger data accuracy obligations under state and federal law. If ChatGPT is producing materially false statements about you or your company, that's not just a reputation problem. It may carry regulatory weight depending on your industry and location.

What This Looks Like in Practice

A Denver-based orthopedic surgery group asked ChatGPT about their practice and received a response that credited a procedure their clinic doesn't perform and listed a physician who had left the group two years earlier. Patients were calling to ask about that specific procedure. The fix wasn't technical. We helped the group publish updated, structured content on their own site and secured coverage in two Colorado health journalism outlets. Within the following training cycle, ChatGPT's responses reflected the current roster and correct specialties.

An early-stage SaaS founder in Austin discovered that ChatGPT described her product as a project management tool when it was actually a financial compliance platform. The confusion traced back to a TechCrunch Startup Battlefield blurb from 2022 that used loose language. Because that single authoritative mention outweighed her own website copy in training weight, the wrong framing stuck. After we helped her earn three accurate, detailed write-ups in fintech publications and updated her structured data, the model's description shifted in subsequent testing to reflect the compliance angle correctly.

A Philadelphia-based commercial contractor found that ChatGPT said almost nothing about his firm at all, defaulting to a generic regional competitor when asked for recommendations. No Wikipedia page, no press coverage, no industry association profile. Eighteen months of steady content and citation work later, including a feature in ENR and consistent profiles on AGC and local business journals, changed that. His firm now appears by name in relevant queries, with accurate descriptions of project scale and geographic focus.

By the Numbers

AI chatbot adoption has moved faster than almost any consumer technology in history. A March 2025 Pew Research survey on how the U.S. public views artificial intelligence found that 75 percent of American adults have heard of ChatGPT specifically, and a growing share report using AI tools to look up information about products, services, and local businesses. That's not a niche audience experimenting in private. Those are real people forming real opinions based on what an AI tells them, and that audience is expanding every quarter.

The accuracy problem isn't theoretical. Researchers publishing on arXiv's information retrieval preprint server have documented retrieval-augmented and base large language models producing factual errors at rates that vary sharply by how well-represented a topic is in training corpora. Entities with thin or inconsistent web footprints, exactly the situation most small businesses and independent professionals are in, see higher error rates than entities with dense, cross-referenced coverage. A 2023 Stanford study cited across multiple arXiv submissions found that LLMs hallucinate at least one factual claim in roughly 27 percent of biographical summaries they generate. If your public web presence is sparse, your odds of landing in that 27 percent go up considerably.

Media behavior is shifting in parallel. The Reuters Institute for the Study of Journalism reported in its 2024 Digital News Report that 22 percent of adults in surveyed markets had used an AI chatbot to get news or background information in the prior month, up from a near-zero baseline in 2022. That behavioral shift matters for reputation because it changes where people form their first impression of you. In 2019 the first impression was a Google results page. Today it's increasingly a paragraph generated by a model that can't cite its sources in real time or flag when it's working from stale data. The window between someone asking ChatGPT about your business and forming a judgment is seconds, not the several clicks a traditional search session might take.

Taken together, these numbers tell a consistent story. Chatbot usage is mainstream. Hallucination rates are non-trivial. And the migration away from link-based search toward conversational AI is accelerating. If you've been treating AI reputation as a future concern rather than a present one, the data say you're already behind the curve. The good news is that the corrective steps, publishing clear authoritative content, earning third-party citations, maintaining consistent entity signals across the web, are the same ones that strengthen your overall digital presence regardless of which AI model is asked about you next.

Another Client Situation

A civil litigation attorney based in Nashville, Tennessee came to us in early 2024 after a prospective client mentioned that ChatGPT had described her as a "general practice" lawyer with no noted specialization. She had spent eight years building a focused plaintiff-side employment discrimination practice, had been quoted in two regional business journals, and maintained a well-trafficked website. The problem was that her site's homepage copy led with broad phrases like "experienced attorney ready to help" rather than specific, factual language about her case history and concentration. ChatGPT, working from training data collected before mid-2023, was pattern-matching to generic attorney language instead of her actual positioning. Over a four-month engagement we rewrote her site's primary pages to lead with concrete facts, secured two new bylined articles in Tennessee legal publications, and got her listed with accurate specialty descriptions in three authoritative bar-adjacent directories. When we re-tested across ChatGPT, Claude, and Perplexity in June 2024, all three correctly identified her as a plaintiff-side employment attorney in Nashville. She reported that two new client inquiries that summer mentioned they had "looked her up with AI" before calling.

Drew Chapin

Drew is the founder of The Discoverability Company. He has spent nearly two decades in go-to-market roles at startup projects and venture-backed companies, is a mentor at the Founder Institute, and a Hustle Fund Venture Fellow. Read more about Drew →

Frequently Asked Questions

How do I find out what ChatGPT says about me?

Open chat.openai.com and ask the same questions your customers or employers would ask. Try multiple variations to get a full picture.

What if ChatGPT says something wrong about me?

You cannot edit its responses directly, but you can influence future outputs by building a stronger, more accurate web presence through press coverage, Wikipedia, and consistent data across authoritative sites.

Should I be worried about what ChatGPT tells people about my brand?

Yes. A growing number of people use ChatGPT for research before making decisions. Monitoring AI search results is now as important as monitoring Google.

How often does ChatGPT update what it knows about a person or business?

ChatGPT's base model updates only when OpenAI releases a new training run, which historically happens every several months to over a year. The GPT-4o model that most users interact with has a knowledge cutoff of early 2024, so anything published after that date won't be reflected unless the user has the browsing plugin enabled. Building authoritative content now still matters, because it positions you well for the next training cycle.

Can I ask ChatGPT to correct something wrong it says about me?

You can't submit a direct correction to the model's weights, but OpenAI does offer a limited personal data removal request process at openai.com. The more practical fix is to publish clear, factual content on high-authority sites so that the next training run has better signal to draw from. Think of it as correcting the record on the open web, not patching the model itself.

Does ChatGPT treat a small local business differently than a well-known brand?

Yes, significantly. A national brand with thousands of press mentions, a Wikipedia page, and broad third-party coverage gets synthesized confidently. A small business with thin online presence, say a single-location plumber in Raleigh, often gets a generic or fabricated response because there isn't enough training signal. That gap is exactly where targeted discoverability work pays off fastest.

Is what ChatGPT says about me considered public information?

ChatGPT's outputs are generated, not retrieved, so they don't map cleanly onto traditional privacy frameworks. That said, the FTC has signaled active interest in AI systems that surface inaccurate personal information, and the IAPP has published guidance noting that AI-generated profiles may trigger data accuracy obligations in certain jurisdictions. If ChatGPT is consistently producing false statements about you, documenting those outputs and consulting a privacy attorney is reasonable.

How often does ChatGPT's training data get updated, and does that affect what it says about me?

ChatGPT's underlying models are retrained periodically, not continuously. GPT-4, released in March 2023, had a knowledge cutoff of September 2021 at launch. That gap means content published after the cutoff simply doesn't exist in the model's memory until a new training run picks it up. The practical implication: if you've done meaningful reputation work in the last year or two, a static ChatGPT session may not reflect it yet. Pairing ChatGPT checks with live-search tools like Perplexity gives you a more complete picture of where you stand right now versus where a model trained on older data thinks you stand.

Ready to take control of your online presence?

Schedule a free consultation to discuss your situation and learn how we can help.