How ChatGPT finds and links a page
OpenAI runs separate agents for separate jobs, and publishers routinely block the wrong one. Blocking training is a legitimate choice; blocking search by accident is the common mistake.
- OAI-SearchBot crawls for ChatGPT search results. OpenAI says sites that block it will not be shown as linked sources in search answers
- GPTBot collects training data. Blocking it does not remove you from search, and allowing it does not get you cited
- ChatGPT-User fetches a page when a person in a conversation asks for it, so it behaves like a visitor rather than a crawler
- Answers without search enabled rely on training data and carry no links, so a missing citation there tells you nothing about your pages
- Citations appear as inline source chips and a sources panel, and OpenAI appends utm_source=chatgpt.com to outbound links, which makes the traffic visible in analytics