ChatGPT draws on two sources when it talks about a brand. The first is what the underlying model absorbed during training, which stops at a fixed date. The second is live retrieval: when ChatGPT searches the web, it reads current pages and can cite them. Your own website is one input among many, and usually a small one. The rest comes from publications, forums, reviews and reference sites that describe you.
For the broader picture across engines, see how AI chooses which brands to mention. This article sets out what OpenAI documents about each route, what independent studies have measured, and what is still unknown. Each section marks which of the three it is drawing on.
What does ChatGPT know about you from training?
A large language model learns from a huge snapshot of text, its training data. That snapshot has a knowledge cutoff: the date after which the model has no built-in knowledge. Anthropic’s documentation explains the idea clearly: a model knows its training material well up to a point, and anything newer has to come from live search.
Documented fact. OpenAI’s GPT-5.5 model page lists a knowledge cutoff of December 1, 2025. OpenAI’s current API models list cutoffs of April 30, 2026 and May 18, 2026. So even the newest models are months behind today.
Reasoning. If your brand launched, rebranded or entered a market after those dates, the model itself has never seen it. If you existed earlier but few pages described you, the model may have seen your name too rarely to connect it with your category. Either way, training alone will not fix a gap quickly. New models ship every few months, and each still stops at a fixed date. Our guide on how a new brand gets into AI answers covers that case in detail.
Where does ChatGPT search get its pages?
Documented fact. OpenAI says ChatGPT search draws on third-party search providers plus content supplied directly by partners, naming Associated Press, Axel Springer, Reuters and The Atlantic. In 2023 Microsoft announced that Bing would be the default search experience in ChatGPT. OpenAI’s current pages do not name one provider, so treat the exact mix as unconfirmed.
OpenAI’s publisher FAQ is plain about eligibility: “Any public website can appear in ChatGPT search”, as long as it does not block OpenAI’s search crawler. There is no special submission and no list to join.
Reasoning. When ChatGPT searches, it typically runs queries, gets back pages, reads some of them and writes an answer, a pattern known as retrieval-augmented generation. A page that is not indexed by the search sources behind it cannot be retrieved. That is why indexing of real, crawlable pages, yours and other people’s, is the precondition for everything else.
Which OpenAI crawlers visit your site, and what does each do?
A crawler is a program that fetches web pages automatically. OpenAI documents three, and they do different jobs. You control them in your robots.txt file.
| Crawler | What OpenAI says it does | What blocking it means |
|---|---|---|
| GPTBot | Collects content to make foundation models more useful and safe | Your pages should not be used to train future models; no effect on ChatGPT search |
| OAI-SearchBot | Surfaces websites in ChatGPT’s search features | Your pages will not be shown in ChatGPT search answers (bar navigational links) |
| ChatGPT-User | Fetches a page when a person’s request in ChatGPT needs it | robots.txt rules may not apply, because a user started the action |
All three rows come from OpenAI’s bot documentation. OpenAI adds that robots.txt changes take about 24 hours to reach its systems. Its FAQ also notes that even a blocked page can appear as a bare link and title in ChatGPT Atlas if OpenAI finds the URL elsewhere.
The practical point: some sites blocked every OpenAI bot when GPTBot was announced and never revisited the setting. If OAI-SearchBot is blocked on your site, ChatGPT search will not show your pages beyond bare navigational links, whatever else you do.
What do studies say about the sources ChatGPT cites?
Industry studies, not vendor facts. These are measurements of ChatGPT’s outputs by third parties. Methods differ, and results shift over time.
- Ahrefs found that in US ChatGPT citations in September 2026, reddit.com had 16.8% of mention share and Wikipedia 7.0%, followed by Consumer Reports, Forbes, YouTube and Walmart.
- Semrush tracked about 230,000 prompts weekly in 2025 and saw Reddit’s presence in ChatGPT answers fall from about 60% to about 10% within weeks. Source mixes can swing sharply.
- McKinsey estimates that brand-owned sites make up roughly 5 to 10% of sources that AI search references. Most of what AI says about you comes from other people’s websites.
- BrightEdge found ChatGPT mentions brands 3.2 times more often than it cites them. A brand mention in the answer is far more common than a citation link.
- Seer Interactive found page-one Google rankings were the strongest predictor of brand mentions in GPT-4o answers, with a correlation of about 0.65. Ahrefs found brand web mentions correlate with ChatGPT visibility at 0.664.
These are correlations, not proof of cause. They do point the same way: brands that are widely described on pages search engines already rank tend to be the brands ChatGPT names.
What can a brand actually control?
Some levers are fully yours. Others you can only influence.
Yours to set:
- Allow OAI-SearchBot in robots.txt. Decide on GPTBot separately; blocking training does not block search.
- Make sure your key pages are indexed and say plainly what you do, for whom and where.
- Track ChatGPT visits. OpenAI says it adds utm_source=chatgpt.com to referral links, so you can filter for it in analytics.
Yours to influence:
- How many independent, indexed pages describe your brand next to your category, use case and market.
- Whether those pages state the same facts: name, what you sell, where you operate, who you serve.
- Whether you appear in the kinds of sources the studies above keep finding: publications, reference pages, reviews and community threads, all used honestly. Our brand mention strategy guide covers the source mix and pacing.
Our guide on why AI doesn’t mention your brand walks through diagnosing which of these is missing.
What is still unknown?
OpenAI does not publish how ChatGPT chooses which retrieved pages to trust, how it weighs one source against another, or which search provider it uses for a given query. Anyone claiming to know those internals is guessing. We do not claim to.
Answers also vary from run to run. In a SparkToro and Gumshoe study, ChatGPT and Google’s AI had less than a 1 in 100 chance of giving the same brand list in any two runs. So the useful measure is how often you appear across many prompts and runs, your mention rate, not a single screenshot. Our guide on measuring AI visibility shows how to set that up.
Suppose a project management tool asks ChatGPT “what are the best tools for small construction firms” ten times. It might be named in three answers, cited in one, and missing from six. That pattern, tracked over weeks, tells you far more than any one answer.
What to do next
- Check robots.txt today: OAI-SearchBot should not be disallowed.
- Run a set of buyer questions through ChatGPT several times and record how often you are named, or get a free AI visibility check from us.
- Once ChatGPT knows you, the next step is being recommended near the top of its answers, which is what The Blue Ocean GPT is built for.
Frequently asked questions
Does ChatGPT get its information from Google?
OpenAI does not say so. It says ChatGPT search uses third-party search providers plus content from news partners, and in 2023 Microsoft announced Bing as the default search experience in ChatGPT. OpenAI's current docs do not name a single provider. In practice, pages that are well indexed by the major search engines are the easiest for any AI search tool to find.
If I block GPTBot, will ChatGPT stop mentioning my brand?
No. GPTBot is OpenAI's training crawler, and blocking it only asks OpenAI not to use your pages for training future models. It does not remove you from ChatGPT search; that is controlled by OAI-SearchBot. Your brand can also be mentioned from other sites that write about you, which no setting on your own site controls.
How can I see if ChatGPT is sending me visitors?
OpenAI says ChatGPT automatically adds utm_source=chatgpt.com to links it sends people to. Filter your analytics by that source to see sessions, pages and conversions from ChatGPT. Remember that ChatGPT names brands far more often than it links to them, so traffic shows only part of the picture. Pair it with regular checks of what ChatGPT says.
How old is ChatGPT's knowledge about my company?
The built-in part is as old as the model's knowledge cutoff, which for OpenAI's current API models falls in spring 2026. Anything newer, such as a launch, a rebrand or a new market, reaches ChatGPT only when it searches the web and finds pages describing it. That is why recent coverage on indexed sites matters so much for new or changing brands.
Can I ask OpenAI to add my brand to ChatGPT?
There is no form for that. OpenAI says any public website can appear in ChatGPT search if it does not block OAI-SearchBot, but there is no documented way to submit a brand for inclusion in answers. The practical route is being described accurately, in context, on sites ChatGPT already reads and cites.
See what AI says about your brand right now.
Tell us your website, your topic and your market in about a minute. We run a free AI visibility check and show you where you stand.
Get my free AI visibility check →