AI assistants that answer questions from the web do it by searching or fetching pages at the time of the question, then citing the pages they drew on. To be eligible, a page has to be reachable by that system's crawler, indexed where the system keeps an index, and clear enough to answer the question being asked. Meeting those conditions makes a page a candidate; it does not guarantee a citation, and each company decides independently what to show.
Two different things: training and retrieval
It helps to separate two ways an AI company might use your pages.
Training is when crawled content may be used to build future models. Retrieval is when an assistant looks something up to answer a specific question now, and can link to the source.
Citations come from retrieval. The major providers now run separate crawlers for each purpose, so a site can make different choices about them. That separation matters later, when you look at your robots.txt.
How does each system find pages?
The table summarises what each company's own documentation says. The links go to those pages.
| System | How pages are found | What the documentation says |
|---|---|---|
| ChatGPT search | OAI-SearchBot | OpenAI says OAI-SearchBot is used to surface websites in ChatGPT's search features, and sites that opt out of it will not be shown in ChatGPT search answers, though they can still appear as navigational links. |
| Claude | Claude-SearchBot and Claude-User | Anthropic says Claude-SearchBot improves search result quality, and Claude-User fetches pages when a user asks a question. Blocking either can reduce a site's visibility in user search. |
| Google AI Overviews and AI Mode | Googlebot and the Google Search index | Google says a page must be indexed and eligible to be shown in Search with a snippet to appear as a supporting link. |
| Microsoft Copilot and Bing | Bingbot and the Bing index | Bing reports which pages are cited in AI answers across Microsoft Copilot, Bing and select partner experiences, in Bing Webmaster Tools. |
ChatGPT
OpenAI documents several crawlers. OAI-SearchBot is the one that matters for appearing in ChatGPT search. GPTBot crawls content that may be used for training, and OpenAI treats the two settings independently, so you can allow OAI-SearchBot while blocking GPTBot. ChatGPT-User visits pages when a user asks ChatGPT to, and OpenAI notes that robots.txt rules may not apply to those user-initiated visits. OpenAI also says robots.txt changes can take about 24 hours to take effect in its systems.
Claude
Anthropic documents three bots. ClaudeBot collects content that may be used for training. Claude-SearchBot indexes content to improve search results. Claude-User retrieves pages in response to a user's question. Anthropic says its bots honour robots.txt, and that disabling Claude-SearchBot or Claude-User reduces what its system can find or retrieve for users.
Google states that the same SEO fundamentals apply to AI Overviews and AI Mode as to Search generally. It also states there are no additional requirements, no special schema.org markup and no AI text file needed to appear. Google describes a "query fan-out" technique, where the system issues several related searches across subtopics, which is why an AI answer can link to a wider set of pages than a classic results page.
Bing and Copilot
Bing's AI Performance report in Bing Webmaster Tools shows which of your pages are cited in AI-generated answers and the grounding queries (the searches the system used to find supporting material) associated with them. It is one of the few places a site owner can see citation activity directly.
What makes a page retrievable?
Access is the first condition. After that, the page has to be a good answer. The pages that tend to work are the ones that:
- state who, what, where and when in plain sentences near the top
- name the company in full and define what it does
- answer a specific question rather than circling a topic
- include original facts, figures or methods that other pages do not have
- keep the important information in text, not only in images or PDFs
- show a publication or update date and stay accurate
This is the same list that makes a page useful to a person, which is not a coincidence. The post on entities, not keywords covers how to write the sentences; the post on your first-party source page covers where the fullest version of your information should live.
Where does a press release fit?
A distributed release puts accurate, dated information about your company on pages beyond your own site. Those pages can be crawled and retrieved like any others, and a release that links back to a strong source page gives a retrieval system a route to your fullest information.
It is still one input among many. An assistant answering "which clinics in Leeds offer home physiotherapy" will draw on whichever pages answer that question most directly, which may be your site, a directory, an independent article or none of them. A release cannot make your site answer questions it does not answer.
What can no one promise?
No publisher, SEO service or distribution network, including media:on, can guarantee that an AI assistant will cite you. The companies running these systems decide what to crawl, index, retrieve and show, and they change how they do it. Bing's own guidance says no tool can guarantee when or how content will appear in AI-generated results, and describes its Citation Share figure as an observational metric, not a ranking.
Two further points are worth keeping in mind:
- A citation is not a ranking. Being cited in one answer says nothing about the next question, and patterns shift as models, user behaviour and the web change.
- A citation is not traffic. Many people read the answer and never click. Treat citations as one more visibility layer, alongside search, referrals and enquiries. The post on measuring press release results covers how to track them together.
Frequently asked questions
If I block GPTBot, will I disappear from ChatGPT search?
Not according to OpenAI's documentation. GPTBot relates to training; OAI-SearchBot relates to ChatGPT search, and the two settings are independent. Blocking OAI-SearchBot is what removes pages from ChatGPT search answers.
Do I need an AI-specific file or schema for Google's AI features?
No. Google states that no new machine-readable files, AI text files or special schema.org markup are needed. The page must be indexed and eligible to show in Search with a snippet.
How can I see whether AI systems cite my site?
Bing Webmaster Tools includes an AI Performance report showing cited pages and grounding queries for Microsoft Copilot, Bing and some partner experiences. For other assistants, look at referral traffic in your analytics and check answers to the questions your customers ask.
Can I check my robots.txt is not blocking these crawlers?
Yes. The post on whether your site is readable by AI walks through robots.txt, the crawler names and the rest of the technical checklist.
media:on Editorial. Written and reviewed by the media:on editorial team, who edit every release before it files.