TL;DR / Key Takeaways
- The 2026 vanity search isn't Google — it's asking ChatGPT, Perplexity, Gemini, Claude and Grok what they know about you.
- Here's how to check for free in ten minutes, why it now matters (AI is the first reader in hiring, dating and due diligence), the honest limits (hallucination, common-name collisions, staleness), and why you can't delete what AI says — you can only influence the sources it reads.
Short answer: To see what AI "knows" about you, just ask the assistants directly — open ChatGPT, Perplexity, Gemini, Claude and Grok and ask "Who is [your name]?" and "What can you tell me about [your name]?" Then note what each one says, what it gets wrong, and whether it confuses you with someone else. A new viral tool called In the Weights scores how strongly the models recall you from training alone. This is the modern "vanity search" — and it now matters because AI is often the first thing that reads you before a recruiter, a date, or an investor ever does.
→ **See exactly what AI says about you and your business**
The AI vanity search, explained
Googling your own name is an old habit. The 2026 version is asking a chatbot what it knows about you — and the results are stranger, because the AI isn't reading a list of links. It's reconstructing you from a blur of training data and, sometimes, live web search. The trend went mainstream in June 2026 when two former OpenAI staffers, Thomas Dimson and Joey Flynn, launched **In the Weights, a site that queries GPT, Gemini, Claude, Grok and Llama with "Who is this person?" and returns a recognition strength score* up to roughly 996. Macaulay Culkin scores 988; a working tech journalist lands around 641 (top 6%). Dimson's framing: "Being in the weights means your existence was deemed important in the process of creating superhuman artificial intelligence."* He added that reception "struck a nerve."
The same nerve powered the earlier "make a caricature of me based on everything you know about me" fad — fun until you realize how much personal data you're handing the model to play along. Whether you find it flattering or creepy, the underlying question is serious: what does the machine actually say about you when someone asks?
How to check what AI says about you (free, in 10 minutes)
You don't need a tool for the first pass. Do this by hand:
- 1Ask each engine directly. In ChatGPT, Perplexity, Gemini, Claude and Grok, run "Who is [your full name]?" and "What is [your name] known for?" Use a logged-out or temporary session so your own chat history doesn't color the answer.
- 2Add your context — "[Name], founder of [company]" or "[Name], [city]" — to see whether the model can disambiguate you from a namesake.
- 3Ask the buyer's question, not just the vanity one: "What's the best tool for [what you do]?" or "Is [your company] legit?" This is what actually gets asked in the wild.
- 4Screenshot the answers and note the sources. Perplexity and Gemini cite links; those citations are the real leverage point, because they show where the model got its picture of you.
- 5Try a footprint scorer like In the Weights to gauge how strongly you exist in the models' baseline memory versus only when they search the live web.
Why it now matters: AI is the first reader
This stopped being vanity the moment AI became the gatekeeper. An estimated 99% of Fortune 500 companies now use AI to filter job applicants (Bricker). 44% of US adults now use ChatGPT (Pew Research, June 2026), and asking an assistant to vet a person or a company before a meeting is now ordinary behavior. Even dating has caught on — about 1 in 7 online daters (14%) use AI to help navigate their dating lives (Singles in America / Kinsey Institute). Recruiters, buyers, investors, journalists and first dates increasingly form their first impression of you through an AI's paraphrase — before they ever reach your website or LinkedIn.
The honest limits: it's often wrong
Here's the part the viral tools gloss over. What AI "knows" about you is frequently incomplete, outdated, or invented:
| Limitation | What it looks like | Why it happens |
|---|---|---|
| Hallucination | Confident, fluent, wrong — a fake job title, a made-up award, a company you never worked at | Models predict plausible text; a thin data footprint gets filled with guesses |
| Common-name collisions | Your bio blended with a stranger who shares your name | The model can't disambiguate when public sources don't clearly separate you |
| Staleness | An old role, a shut-down startup, a years-out-of-date fact | Training data and cached sources lag the real world by months |
| Silence | "I don't have information about that person" | You have too little consistent third-party coverage to register at all |
Only 29% of Americans say they trust AI output even as usage climbs (Pew) — but "distrust in the abstract" doesn't stop people acting on a confident answer in the moment. If the model invents a detail about you, the reader usually can't tell. That's the risk, and it's why measuring is only step one. (For the business version of this problem, see AI hallucinations about your brand.)
You can't delete it — you influence the sources
The uncomfortable truth: there's no "edit my profile" button inside ChatGPT, no delete key, no form to submit a correction. AI answers are reconstructed from the public web and training data, so the only real lever is to change the sources the models read from. That work is slow and unglamorous, but it's the only thing that works:
- 1Make your facts consistent everywhere — your site, LinkedIn, Crunchbase, bios, profiles all telling one coherent story. Contradictions are exactly what make models hallucinate or merge you with a namesake.
- 2Earn credible third-party coverage — models echo what other trustworthy sources say. Real press, a genuinely warranted Wikipedia entry, respected community threads and reviews shift the picture more than anything on your own domain.
- 3Publish clear, answer-shaped pages that directly state who you are and what you do, with facts a model can lift verbatim.
- 4Stay crawlable — if you block GPTBot, ClaudeBot or Google-Extended, you've opted out of being read at all.
Anyone promising to "submit you to ChatGPT" or "guarantee" what AI says is selling something that doesn't exist — the full breakdown is in the pillar, AI Reputation Management in 2026. And if you sell a product, the sharper question is whether the models recommend you: does ChatGPT recommend your product?
Enjoying this? Get one like it in your inbox each morning.
one email a day · unsubscribe in two clicks · no third-party tracking
Where Stork fits
A manual vanity search is a great start, but it's five browser tabs, no history, and no structure. Stork's AI Reputation Report ($29, one-time) does the measurement half properly: it runs your real buyer questions live across ChatGPT, Perplexity, Gemini, Claude and Grok, captures the verbatim answers, flags where they're wrong or blank, shows who AI names instead of you, lists the sources each engine cites, and hands you a prioritized to-do list of the pages to influence. No guarantees, no "we'll get you into ChatGPT" — just the truth, and the honest next step.
→ **Run your AI vanity search the thorough way**
_Related reading: AI Reputation Management in 2026 (the honest map of the whole category), does ChatGPT recommend your product?, and AI hallucinations about your brand._
Frequently asked questions
How do I check what ChatGPT knows about me?
Open ChatGPT in a logged-out or temporary chat (so your own history doesn't bias it) and ask "Who is [your full name]?" and "What is [your name] known for?" Add context like your company or city to help it disambiguate. Repeat in Perplexity, Gemini, Claude and Grok, and note where each one is wrong, outdated, or confusing you with someone else.
What is an AI vanity search?
It's the 2026 version of googling yourself: asking AI assistants what they know about you instead of scanning search results. Tools like In the Weights formalize it by scoring how strongly the major models recall you from their training data alone. It matters because AI is increasingly the first thing that reads you before a human does.
Can I delete or correct what AI says about me?
Not directly — there's no edit button or submission form inside ChatGPT, Perplexity or Claude. AI answers are reconstructed from the public web and training data, so the only reliable fix is to influence those sources: make your facts consistent everywhere, earn credible third-party coverage, publish clear pages, and stay crawlable. It's slow, and no one can guarantee a result.
Why does AI get facts about me wrong?
Because language models predict plausible text rather than look you up in a database. With a thin or contradictory public footprint, they fill gaps with confident guesses (hallucination), blend you with a namesake (common-name collision), or repeat stale data. That's why an AI answer can sound authoritative and still be flatly false.
Does In the Weights show everything AI knows about me?
No. It measures how strongly the models recall you from training alone — a recognition score — not the full picture a live, web-searching assistant would assemble for a recruiter or buyer. For that you need to run the actual buyer questions across multiple engines and read the citations, which is what a structured AI reputation report does.
Disclosure: Stork sells a $29 AI Reputation Report and runs an AI-tools directory. This article exists because the honest version of "what does AI know about me" was missing — we'd rather show you how to check for free and tell you the real limits than sell you a guarantee no one can keep.