ChatGPT-User
Fetches a page when a ChatGPT or custom GPT user asks for it. OpenAI says robots.txt may not apply, and it does not decide what appears in ChatGPT search.
| Operator | OpenAI |
|---|---|
| Type | AI: fetch on a user's request |
| robots.txt token | ChatGPT-User |
| Follows robots.txt | Partly; Crawl-delay not documented |
| Sends visitors | Indirectly |
| Links to your pages | Sometimes |
| Can end up in AI training | Not documented |
| Verification |
IP list
Our tracker checks the IP list. |
| Our call |
Publisher: Allow E-commerce: Allow Expert site: Allow What the profiles mean |
How to recognize it
User agent, as it reaches our logs (identical to OpenAI’s docs):
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; ChatGPT-User/1.0; +https://openai.com/bot
Match on the token ChatGPT-User. On our sites 98% of requests with this user agent were fake, so always check the IP.
- robots.txt token:
ChatGPT-User - IP list (JSON): chatgpt-user.json
- Reverse DNS: none published by OpenAI.
What it does
ChatGPT-User opens a page when a person using ChatGPT or a custom GPT asks for something that needs it. OpenAI says it does not crawl the web automatically and is not used to decide whether content appears in ChatGPT search.
What you gain by allowing it
- ChatGPT can read your page when a user points it there or when an answer needs it. The answer may cite and link the page.
What you lose
By allowing it
- Your content is read and summarised for a user, who may not click through.
- OpenAI’s page does not say whether these fetches are used for training.
By blocking it
- ChatGPT cannot open your page for its users, so it answers without your content or from older copies.
- robots.txt may not be enough. OpenAI says the rules may not apply to user-initiated actions. A firewall rule on the IP list is the reliable way to block it.
How to verify it
Match the IP against chatgpt-user.json; the one-liner on the GPTBot card works with this URL. Cloudflare flags the real fetcher as a verified bot in the category “AI Assistant”, and in our data the flag matched the list on every request.
Gotchas and history
- Counting ChatGPT-User by user agent measures the scanner, not ChatGPT. On our sites the user agent count was 43 times the verified one.
- Its name says nothing about ChatGPT search. Pages are found for search by OAI-SearchBot.
What we observed
Data: Cloudflare Logpush for two of our sites, 19-26 October 2026 (8 days).
- 782 requests carried a ChatGPT-User user agent:
- 18 (2%) came from OpenAI’s list, all from Microsoft’s network (AS8075);
- 756 of the 764 fakes came from the Google Cloud scanner;
- Cloudflare’s verified-bot flag matched the list on all 782.
- The 18 verified requests fetched 13 different pages on both sites, a homepage and articles, spread over all 8 days. None asked for
/robots.txt. 17 got a 200.
About OpenAI
- GPTBot collects content for training foundation models. Disallowing it opts the site out of training only.
- OAI-SearchBot finds pages for search results in ChatGPT. Disallowing it removes the site from those results. OpenAI gives about 24 hours for a robots.txt change to take effect there.
- ChatGPT-User opens pages when a user asks ChatGPT or a custom GPT to. It does not crawl on its own, and OpenAI says robots.txt rules may not apply to it.
- OAI-AdsBot visits only pages submitted as ads, to check the landing page.
OpenAI states that each setting is independent. When a site allows both GPTBot and OAI-SearchBot, OpenAI may use one crawl for both. The common mix-up is blocking GPTBot to stay out of ChatGPT: that only stops training, while search visibility depends on OAI-SearchBot.