ChatGPT-User
Fetches a page in real time when a ChatGPT user or agent follows a link. OpenAI states robots.txt may not apply to it.
Reference reviewed by Sona
| Operator | OpenAI |
|---|---|
| Powers | Live page fetches when ChatGPT acts for a user |
| Purpose | User-triggered fetch |
| User-agent token | ChatGPT-User |
| Respects robots.txt | No (user-triggered) |
ChatGPT-User is not a bulk crawler. It fires when a person using ChatGPT triggers a fetch - asking ChatGPT to open or summarize a specific URL, or letting an agent browse on their behalf. OpenAI describes it as "not used for crawling the web in an automatic fashion."
The important change is its robots.txt status. OpenAI's current crawler documentation states that because these visits are initiated by a user, robots.txt rules may not apply to ChatGPT-User. Earlier guidance was read as a straightforward promise of compliance, and a lot of published robots.txt advice still assumes it. Treat a Disallow for this token as a request rather than a control.
Independent measurement supports the weaker reading. TollBit's State of the Bots reporting for the first half of 2026 found ChatGPT-User reaching pages that had explicitly disallowed it on roughly half of the European sites that named it - the widest reach of any agent in that group. If your reason for blocking is a legal or contractual obligation rather than a preference, robots.txt alone will not satisfy it.
That leaves a genuine judgment call rather than a technical one. These are the highest-intent visits any AI system makes at your site: a specific human asked for your specific page. Blocking them degrades the answer that person receives about you. Most sites allow ChatGPT-User deliberately, and enforce at the edge only when they have a concrete reason.
How ChatGPT-User behaves
- One-off fetches of specific URLs, typically a single page rather than a crawl of your site.
- Arrives with no warm-up - a page can go from zero bot traffic to a fetch the moment a user pastes its URL into ChatGPT.
- May reach pages your robots.txt disallows; OpenAI documents robots.txt as possibly not applying to user-initiated visits.
- Agent and connector browsing inside ChatGPT can surface under this token too, so volume rises with agentic usage rather than with your publishing cadence.
Full user-agent string
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; ChatGPT-User/1.0; +https://openai.com/bot
Allow ChatGPT-User
A user explicitly asking ChatGPT about your page gets an answer grounded in what the page actually says - the most intent-heavy AI visit you receive.
User-agent: ChatGPT-User Allow: /
Block ChatGPT-User
You need to prevent all programmatic access from ChatGPT, including user-initiated fetches - but enforce it at the WAF, because robots.txt is documented as possibly not applying.
User-agent: ChatGPT-User Disallow: /
Heads up: ChatGPT-User does not reliably honor robots.txt. To enforce a block, combine the rule above with server- or firewall-level filtering of the user-agent.
How to verify ChatGPT-User
Published IP range feed
OpenAI publishes a separate feed at openai.com/chatgpt-user.json. Verification matters more for this token than for the others: because it is the one that reaches pages regardless of robots.txt, it is the identity scrapers most want to borrow, and traffic arriving from outside the published ranges has no claim to be user-initiated. If you enforce a block, enforce it on verified IPs and drop the rest as forgeries.
Check an IP against this botCommonly confused with ChatGPT-User
GPTBot honors a Disallow; ChatGPT-User is documented as possibly ignoring one. Blocking OpenAI training is achievable via robots.txt, blocking user fetches generally is not.
OAI-SearchBot indexes ahead of time so ChatGPT can suggest your page; ChatGPT-User then fetches it live. They are two halves of one user journey.
The same pattern at a different vendor - Perplexity-User is likewise documented as generally ignoring robots.txt because a person initiated the request.
ChatGPT-User FAQs
Does ChatGPT-User respect robots.txt?
Not reliably. OpenAI's crawler documentation states that because these fetches are initiated by a user, robots.txt rules may not apply. Independent measurement in 2026 found it reaching disallowed pages on about half of the European sites that had listed it.
How do I actually stop ChatGPT-User?
Filter the user-agent at your server, CDN, or WAF, and scope the rule to the IP ranges in openai.com/chatgpt-user.json so you are blocking the real bot rather than anything wearing its name. A robots.txt Disallow on its own is not an enforcement mechanism.
Is blocking ChatGPT-User a good idea?
Usually not. These fetches happen because a specific person asked about your specific page, which makes them the highest-intent AI traffic you get. Blocking them means ChatGPT answers that question about you from stale or second-hand information instead.
Why did ChatGPT-User hit one page on my site with no other bot traffic?
That is the expected pattern. It is not a crawler working through your sitemap - it fetched the one URL a user named in a conversation, so isolated single-page hits are normal.
Can ChatGPT-User read your page right now?
Test any URL and see exactly what AI crawlers receive.