ChatGPT-User

This is the fetch that happens while somebody is waiting. A person asked ChatGPT something, the model decided your page was worth reading, and it went and read it — so a hit here is the closest thing to a referral that AI traffic produces. It usually arrives as a single request for one URL, not as a crawl.

OpenAI runs four separate crawlers and gives each one its own token, which is more considerate than it first appears: it means you can refuse the training crawl and keep the one that puts you in ChatGPT's answers, using two lines in a robots file. Most operators on this site do not offer that choice.

On this site ChatGPT-User is counted as an AI answer, which is the row it appears in on your dashboard. The category matters because it decides which number moves when this crawler visits, and the four are reported separately for exactly that reason.

One request, one URL, no crawl. It arrives because a person asked ChatGPT something in that moment, so the pattern follows human hours rather than a schedule, and the pages hit are the ones people are actually asking about — which makes this the most interesting row on the panel to read by page.

Checked as ip ranges + reverse dns. The user agent has to carry the token, and then the address is compared against what OpenAI publishes — refetched every six hours, so a newly announced range is picked up without a deploy.

A request that matches neither is recorded as unverified rather than as a forgery: a published list can be incomplete, and an accusation needs evidence. Only a reverse DNS record resolving to somebody else produces a spoofed verdict, which is proof rather than the absence of it.

OpenAI's assistant can no longer open your page to answer a question about it. People will still be told about you from whatever the model already knows, but without the current page, a link, or a citation — which is usually the opposite of what a site owner wants.

This is the crawler that brings people back. When an assistant fetches a page to answer with, it usually shows the source and links to it, so the visit can turn into a reader. Blocking it removes you from the answer without removing the question.

The argument for blocking is the same one that applies to any zero-click surface: if a model can answer from your page, some readers get what they came for without visiting. That matters most if your pages are the product — reference material, calculators, documentation — and least if they exist to sell something the model cannot deliver.

OpenAI sends 5 crawlers and gives each its own token, so they can be allowed and refused separately. The tokens differ by a few characters and what the fetch means does not:

ChatGPT-Agent — ai assistant. ChatGPT's agent mode browsing on a person's behalf. OAI-SearchBot — search indexing. Builds the index behind ChatGPT search. GPTBot — ai training. Collects public pages that may improve future OpenAI models. OAI-AdsBot — other bots. Checks pages submitted as ChatGPT ads.

Confusing two of these is the most common mistake with this operator, and it is expensive in one direction: a rule meant for the training crawler that lands on the search crawler removes you from results without stopping any training.