ChatGPT, Claude, Gemini, Perplexity: Which AI Uses Which Sources?
ChatGPT, Claude, Gemini and Perplexity find sources on the web in different ways: ChatGPT uses OAI-SearchBot for its search features, Claude uses Claude-SearchBot and Perplexity uses PerplexityBot, while Google's AI features rely on the Google Search index. The criteria for which sites they feature are not explained in detail; what is documented are the conditions for access and indexing.
There is no single answer to "which sites does AI cite as sources?", because the four platforms do not use the same system. Each has its own crawler, its own controls and its own rules.
This article compares how AI search engines retrieve sources from the web, based on official documentation, and explains what to check so that your site can appear as a source on these four platforms.
The four platforms at a glance
ChatGPT | Claude | Google (AI Overviews, AI Mode, Gemini) | Perplexity | |
|---|---|---|---|---|
Bot that indexes pages for search | OAI-SearchBot | Claude-SearchBot | Googlebot (Google Search index) | PerplexityBot |
Bot that opens pages on user request | ChatGPT-User | Claude-User | Results come from the Search index | Perplexity-User |
robots.txt for user-initiated bot | According to OpenAI, may not apply | Site owners can control access via Claude-User | Googlebot rules apply | According to Perplexity, generally ignored |
Separate bot for training | GPTBot | ClaudeBot | Google-Extended (for training and grounding in other systems) | None; Perplexity states PerplexityBot is not used for foundation model training |
If you block the search bot | You are not shown in ChatGPT search answers, but may appear as a link | Visibility and accuracy in search may decrease | Blocking Googlebot removes you from Search | You are removed from Perplexity's index |
Information is taken from the official documentation of each company.
How does ChatGPT find sources?
OpenAI defines three separate user agents, each controlled independently. OAI-SearchBot is used to surface sites in ChatGPT's search features; sites that block it are not shown in ChatGPT search answers, though they can still appear as navigational links. GPTBot is for training generative AI models. ChatGPT-User is used when a user asks a question and ChatGPT opens a page; OpenAI says this is triggered by user action and that robots.txt rules may not apply in that case. OpenAI recommends managing search visibility through OAI-SearchBot in robots.txt. It also says robots.txt changes for search can take about 24 hours to be reflected in its systems.
Source: OpenAI, Overview of OpenAI Crawlers.
How does Claude find sources? (Claude SEO)
Anthropic also defines three bots. ClaudeBot collects content that may contribute to model training. Claude-User is used to open sites when Claude users ask a question; blocking it may reduce visibility in user-directed web search. Claude-SearchBot crawls the web to improve the quality of search results; if it is blocked, your content is no longer indexed for search, and visibility and accuracy in search results may decrease. Anthropic states that its bots respect robots.txt directives.
Source: Anthropic, Does Anthropic crawl data from the web.
How do Gemini and Google's AI features find sources? (Gemini SEO)
According to Google's guide, AI Overviews and AI Mode are built on Google's core ranking and quality systems; to improve responses, they retrieve relevant pages from the Search index (RAG or grounding) and run several related queries (query fan-out). To be shown as a supporting link in these features, a page must be indexed and eligible to appear in Google Search with a snippet. The guide says there are no additional technical requirements or special optimisation needed for AI Overviews and AI Mode; it also says that, from Google's perspective, optimising for AI search under terms such as "AEO" and "GEO" is optimising for the search experience, in other words SEO. The same guide says a site must be included in generative AI features in Search Console to appear in them; check this in your Search Console settings.
If you want to limit training and grounding in Google's other systems, such as the Gemini app, there is the Google-Extended control. We could not find further detail in the official documentation on how the Gemini app selects sources, so we make no claim about it.
Sources: Google Search Central, AI features and your website, Google Search Central, Optimizing your website for generative AI features.
How does Perplexity find sources? (Perplexity SEO)
Perplexity's PerplexityBot is designed to surface and link websites in Perplexity search results; Perplexity says it is not used for foundation model training and recommends allowing it in robots.txt. Perplexity-User is the agent that opens a page when a user asks a question; because the user requested it, it generally ignores robots.txt rules. Perplexity states that it will not crawl the full or partial text of a site that blocks PerplexityBot in robots.txt, but may still index the domain, headline and a brief factual summary. It also says it works with third-party crawlers for its search index and has made adjustments with these providers so that robots.txt is respected.
Sources: Perplexity, PerplexityBot, Perplexity, robots.txt FAQ.
Which sites does AI cite as sources?
The honest answer: no platform explains its source selection criteria in detail, and there is no verifiable official data on which sites they feature more often. So treat claims like "these kinds of sites become sources" with caution. What the official documentation has in common is this:
The page must be in the relevant platform's index, which requires access for its search bot.
On Google, the page must be indexed and eligible to be shown with a snippet.
Google says unique, helpful, people-first content will have the most impact in the long run.
None of these guarantees that you will be cited; they only meet the preconditions.
robots.txt example for being a source on all four platforms
Allowing search bots is independent of your decision about training bots. The example below shows only the permissions needed for search and citation; decide separately on training bots (GPTBot, ClaudeBot, Google-Extended), and get the view of your legal and brand teams on that decision.
User-agent: OAI-SearchBot Allow: / User-agent: Claude-SearchBot Allow: / User-agent: PerplexityBot Allow: / User-agent: Googlebot Allow: /Even if robots.txt allows access, CDN or firewall rules can stop bots; check both layers. Extra files such as llms.txt do not replace this access, and Google says it does not use such a file for Search.
What should you do to become a source in AI search engines?
Verify access. Check robots.txt and CDN rules for the search bots above.
Write direct answers. Put a 40–60 word paragraph answering the main question at the top of every important page.
Keep information consistent. Your company description, services and certifications should be the same on your site and your public profiles.
Match structured data to the page. Schema should reflect the visible text; Google says it is not required but should match the page.
Test on all four platforms. Ask buyer questions in ChatGPT, Claude, Gemini and Perplexity, record the results and repeat monthly. For how to do this in 30 minutes, see Why Isn't Your Brand Showing Up in ChatGPT?; for the conceptual framework, see What Is GEO? and SEO vs GEO vs AEO.
Clockwork's view from the field
Clockwork is an independent digital and creative agency based in Istanbul with 25 years of experience, certified to ISO 27001 and ISO 12207.
Frequently asked questions
Which sites does AI cite as sources?
The platforms do not explain their source selection criteria in detail. What is documented is that the page must be in the platform's index and access for the relevant search bot must be open. No platform guarantees that you will be cited.
Do ChatGPT, Claude, Gemini and Perplexity use the same sources?
No. ChatGPT uses OAI-SearchBot for its search features, Claude uses Claude-SearchBot and Perplexity uses PerplexityBot, while Google's AI features rely on the Google Search index. Each platform has its own crawler and controls.
If I block a training bot, will I lose visibility in AI search?
Generally no, because the settings are separate. At OpenAI, GPTBot and OAI-SearchBot are independent; at Anthropic, ClaudeBot and Claude-SearchBot are separate. Perplexity states that PerplexityBot is not used for model training.
Is Gemini SEO a separate effort?
According to Google, optimising for generative AI features in Search is optimising for the search experience, in other words SEO; no special optimisation is needed for AI Overviews and AI Mode. The guide also says the site must be included in generative AI features in Search Console.
Conclusion
The four platforms use four different routes: ChatGPT, Claude and Perplexity use their own search bots, and Google uses its Search index. What they all require is access and indexing; being cited cannot be guaranteed. Verify access first, then strengthen your content and the consistency of your information, and test regularly.
Clockwork is an independent digital and creative agency based in Istanbul with 25 years of experience. If you would like to assess your site's AI search visibility together, explore our services or get in touch.