Does ChatGPT cite your website?

Updated 2026-06-13 · maintained by Unsourced

Most advice tells you to “block the AI bots.” For ChatGPT that's usually a mistake, and it comes from confusing three crawlers that do three different jobs. Block the wrong one and you won't protect your content. You'll quietly remove yourself from ChatGPT's answers. Here's what's actually true, and how to find out whether ChatGPT cites you instead of guessing.

What most guides get wrong

Being cited and being crawled are not the same thing. ChatGPT can name your site from training data it learned months ago without ever visiting, or link to you live from a fresh search. In the answer they look identical. For your traffic they're worlds apart, and most tools can't tell which one happened. That gap is why we built Origin Signal.
The training bot and the search bot are different bots. Blocking GPTBot won't remove you from live answers. Blocking OAI-SearchBot will. Any guide that treats “OpenAI's bot” as one thing has it wrong.
A log line isn't proof. Anything can write “GPTBot” into its user-agent. Until a request is checked against OpenAI's published IP ranges and reverse DNS, it's just a claim, and it might be an impostor using the name.

How ChatGPT surfaces sources

ChatGPT answers from training data, from a live web search, or by fetching a page you point it to. The live answers are the ones that carry clickable citations linking back to the source, and those links can send you real traffic. They're powered by OAI-SearchBot, backed by OpenAI's partnership with Microsoft's Bing. Like almost every AI crawler it doesn't run JavaScript, so the version of your page in plain server-rendered HTML is the version it can read.

The crawlers behind ChatGPT

OAI-SearchBot surfaces and links your pages in ChatGPT Search. Block it and you disappear from live answers.
ChatGPT-User fetches a page in real time when someone asks ChatGPT to read it. Block it and users can't open your page through ChatGPT.
GPTBot gathers data that may train future models. Block it and you're less likely to be learned by the next model, which can reduce mentions later, but it changes nothing about live search.

OpenAI keeps these controls separate on purpose, so you can stay visible in search while opting out of training. All three obey robots.txt. Just remember robots.txt is a request honest bots follow, not a wall that stops an impostor.

How to check whether ChatGPT cites you

Typing your questions into ChatGPT and watching for your name is a reasonable gut-check, but its answers shift from session to session, so a single look proves nothing. Unsourced works the opposite way to tools that blast a generic prompt list and hope. We build the test questions from your own signals: the searches you already rank for in Google and Bing Search Console, the topics in your published pages, and the queries that bring real visitors to your site, phrased the way someone would actually ask an assistant. We run them on a schedule, log every time ChatGPT cites your domain and which competitors it picks instead, and use Origin Signal to tell a live-search citation apart from one pulled from older training data. That's evidence built on your real demand, not a canned checklist.

How to improve your chances

Let the right bot in. The one that controls search visibility is OAI-SearchBot, not GPTBot. Make sure it isn't blocked.
Keep the answer in your HTML. The crawler doesn't run JavaScript, so anything that appears only after scripts load won't be read.
Write the way people ask. Pages that answer a specific question directly are easier to surface than long, unfocused ones.
Stay current, and stay in Bing. Freshness helps, and the OpenAI partnership means a Bing presence can help you appear.
Verify before you trust, then measure. Confirm the crawls hitting you are genuinely OpenAI's, and track whether citations actually follow.

Nobody can promise you a ChatGPT citation, and anyone who does is guessing. What you can control is real: unblock the crawler that matters, publish answers worth citing, and measure what actually happens. Doing that last part honestly is the whole reason Unsourced exists.

Common questions about ChatGPT

Does ChatGPT cite sources with links?

Yes. When ChatGPT answers from a live web search it shows clickable inline citations that link back to the source, which can send referral traffic. Answers drawn purely from training data may mention a site without a live link.

Will blocking GPTBot remove me from ChatGPT?

No. GPTBot is the training crawler. The crawler that controls whether you appear in ChatGPT's live answers is OAI-SearchBot. Blocking GPTBot affects training, not search visibility.

How do I know if ChatGPT is citing my site?

Asking ChatGPT directly is anecdotal and changes between sessions. Unsourced measures it over time by testing the queries you actually rank for, recording when your domain is cited, and showing whether the mention came from a live crawl or training data.

Other AI assistants

Is ChatGPT citing your site, or your competitors?

Unsourced tests the queries you already rank for, records who gets cited, and shows whether each mention came from a live crawl or training data.

See whether ChatGPT cites your site →

14-day free trial · no card required · cancel anytime