ChatGPT-User
Definition: ChatGPT-User is the user agent OpenAI uses when ChatGPT or a custom GPT visits a web page because a person asked it to, rather than to crawl the web automatically.
Also called: ChatGPT User agent
Updated
User-triggered, not a crawler
ChatGPT-User shows up when a person in ChatGPT asks about a page, pastes a link, or uses a custom GPT that fetches web content. OpenAI’s crawler documentation makes three points about it:
- It is not used to crawl the web automatically.
- It is not used to decide whether content may appear in ChatGPT search; that is OAI-SearchBot’s role.
- Because these visits are initiated by a user, robots.txt rules may not apply.
Other providers take different positions. Perplexity says its user-triggered fetcher, Perplexity-User, generally ignores robots.txt. Anthropic says its Claude-User agent honors it.
What its visits tell you
A ChatGPT-User request in your server logs means a real person asked ChatGPT about that specific URL. That is a demand signal you will not see in your analytics, because the person may never click through. Counting these requests by page shows which content people are taking into AI conversations.
grep "ChatGPT-User" access.log | awk '{print $7}' | sort | uniq -c | sort -rn | headShould you block it?
Our view: for most sites, no. Each request stands for a person who wanted your page. Blocking it means ChatGPT answers from other sources, or from memory, when someone asks about you specifically.
If you do need to block it, for example on a members-only section, robots.txt may not be enough. Use authentication, or a server or firewall rule matched against the user agent and OpenAI’s published IP ranges. The AI crawlers guide compares how every major user-triggered agent treats robots.txt.