Check
ChatGPT-User: what it is and whether to allow it
Fetches a page when a person asks ChatGPT to look at it. This is the closest thing to a real visit on the crawling side — it happens because a human asked.
Test whether you are blocking it
curl -sS -A "ChatGPT-User/1.0" -o /dev/null -w "%{http_code} %{size_download}\n" https://example.com/
A 403 means a firewall or bot rule matched. A tiny body with status 200 means the crawler
reached a shell or a challenge page and read nothing useful. Both are failures; they just look
different.
Should you allow it
If ChatGPT-User is refused, a user who pasted your URL gets an error or a summary from memory instead of your actual page. That is a visible failure, not a silent one.
Where this fits
Crawler access is check A1 of the fifteen-point audit, and it comes first: nothing downstream matters if the request is refused. The full list is in the standard; the sequence that avoids redoing work is in the repair guide.