Glossary
OpenAI openai.com ChatGPT-User OpenAI's agent for user-triggered page visits.
The OpenAI agent that reads your page when a ChatGPT user asks about it, not on a crawl schedule. Prefer's Agent Analytics shows these visits in your server logs, bot by bot.
ChatGPT-User is the OpenAI user agent that visits a web page when a ChatGPT or Custom GPT user's question needs it; it does not crawl automatically. Prefer's Agent Analytics shows its visits.
Key facts
At a glance.
The entity facts an assistant lifts first, each checked against OpenAI's own documentation.
- Operator
- OpenAI · openai.com
- Type
- User-triggered agent (ChatGPT, Custom GPTs and GPT Actions)
- User agent token
- ChatGPT-User
- Full UA string
- Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; ChatGPT-User/1.0; +https://openai.com/bot
- Crawls automatically
- No. OpenAI says it is not used for automatic crawling
- Respects robots.txt
- Partly. OpenAI says robots.txt rules may not apply
- IP ranges
- Published at openai.com/chatgpt-user.json
- First announced
- Not documented
- Affects ChatGPT search inclusion
- No. OpenAI points to OAI-SearchBot for search opt-outs
How it works
What happens when ChatGPT-User visits.
Three steps, and the first is what makes it different: a person starts it, not a crawl schedule.
- 01
A user asks
Someone asks ChatGPT or a Custom GPT a question, or uses a GPT Action, and answering it needs a web page.
- 02
ChatGPT visits the page
ChatGPT may visit the page with the ChatGPT-User agent. Because the visit is user-initiated, OpenAI says robots.txt rules may not apply.
- 03
The answer uses what it read
The page is read for that user's request. OpenAI says ChatGPT-User is not used to decide whether content may appear in ChatGPT search.
The OpenAI crawler family
ChatGPT-User is one of four. Each has its own job.
OpenAI says each setting is independent, so a rule for one token does not change the others.
Purpose in OpenAI's words, what blocking changes, and whether it decides ChatGPT search inclusion.
| Bot | Purpose | Blocking it means | Decides ChatGPT search inclusion |
|---|---|---|---|
GPTBot | Crawls content that may be used to train OpenAI's generative AI foundation models. | Signals your content should not be used in training. | ✕ No |
OAI-SearchBot | Surfaces websites in search results in ChatGPT's search features. | Your site is not shown in ChatGPT search answers, though it can still appear as a navigational link. | ✓ Yes |
ChatGPT-UserThis term | Visits a page when a ChatGPT or Custom GPT user's question needs it. Not used for automatic crawling. | OpenAI says robots.txt rules may not apply to these user-initiated visits. | ✕ No |
OAI-AdsBot | Validates the safety of web pages submitted as ads on ChatGPT. | Only affects pages you submit as ads. Its data is not used for training. | ✕ No |
Why it matters for AEO
Allow or block: the decision in one sentence.
Should you block ChatGPT-User?
For most brands, no, and Prefer's Agent Analytics shows how often ChatGPT-User actually reaches you. It only visits when a ChatGPT user's question needs your page, OpenAI says robots.txt may not apply to it, and it does not decide whether you appear in ChatGPT search. Blocking it mainly risks a person asking about your page getting nothing back.
- You want ChatGPT users who ask about your pages to get your content
- Your pages are public marketing, docs or editorial
- Your pages answer questions people ask ChatGPT about you
- A section is private or licensed (protect it with a login, not only robots.txt)
- You accept that a Disallow rule here may not be honoured
- You still leave OAI-SearchBot allowed if you want ChatGPT search placement
To manage ChatGPT search opt-outs, OpenAI points to OAI-SearchBot, not ChatGPT-User.
Allow or block it
Three robots.txt patterns and one caveat.
Each names the ChatGPT-User token explicitly. OpenAI says robots.txt rules may not apply to user-initiated visits, so treat a Disallow here as a request.
robots.txtCopy the pattern that matches your decision above.
User-agent: ChatGPT-User
Allow: /
User-agent: OAI-SearchBot
Allow: /
The pattern most brands want: ChatGPT can read your page when asked, and you stay eligible for ChatGPT search.
- Search is a different tokenBlocking ChatGPT-User does not opt you out of ChatGPT search. OpenAI says to use OAI-SearchBot for that.
- Check for a wildcardA User-agent: * group with Disallow: / also covers ChatGPT-User unless a group names ChatGPT-User directly.
- robots.txt is not access controlBecause robots.txt may not apply to these visits, use authentication for content that must stay private.
# Request only: OpenAI says robots.txt rules
# may not apply to user-initiated visits
User-agent: ChatGPT-User
Disallow: /
OpenAI says robots.txt may not apply to these user-initiated visits, so this is a signal, not a lock.
- Search is a different tokenBlocking ChatGPT-User does not opt you out of ChatGPT search. OpenAI says to use OAI-SearchBot for that.
- Check for a wildcardA User-agent: * group with Disallow: / also covers ChatGPT-User unless a group names ChatGPT-User directly.
- robots.txt is not access controlBecause robots.txt may not apply to these visits, use authentication for content that must stay private.
# Request only for user-initiated visits
User-agent: ChatGPT-User
Allow: /
Disallow: /members/
Keep public pages readable, ask it to skip a private section. Protect truly private content with a login.
- Search is a different tokenBlocking ChatGPT-User does not opt you out of ChatGPT search. OpenAI says to use OAI-SearchBot for that.
- Check for a wildcardA User-agent: * group with Disallow: / also covers ChatGPT-User unless a group names ChatGPT-User directly.
- robots.txt is not access controlBecause robots.txt may not apply to these visits, use authentication for content that must stay private.
Verify a visit
How to confirm it was really ChatGPT-User.
Any script can claim the ChatGPT-User user agent. OpenAI documents two things you can check.
- 01
Match the user agent
Look for the ChatGPT-User token in the user agent field. The full string OpenAI publishes ends with a link to openai.com/bot.
- 02
Match the IP range
Compare the request IP against the ranges OpenAI publishes at openai.com/chatgpt-user.json. A ChatGPT-User user agent from an IP outside those ranges is not OpenAI.
203.0.113.42 - - [01/Oct/2026:09:14:07 +0000] "GET /pricing/ HTTP/1.1" 200 18422 "-" "Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; ChatGPT-User/1.0; +https://openai.com/bot"In context
The term in a sentence.
01"Our logs show ChatGPT-User hits on the pricing page, so people are asking ChatGPT about our plans by name."
02"We left ChatGPT-User allowed and kept OAI-SearchBot allowed too, because only OAI-SearchBot decides ChatGPT search inclusion."
Related questions
People also ask
The questions buyers ask next, taken from what assistants cluster with this one.
Is ChatGPT-User a crawler?
Not in the usual sense, and Prefer's Agent Analytics reports it separately from crawlers like OAI-SearchBot. OpenAI says ChatGPT-User is not used for crawling the web in an automatic fashion; it visits a page when a ChatGPT or Custom GPT user's question needs it.
See the crawler comparison →What is the ChatGPT-User user agent string?
Prefer's Agent Analytics spots ChatGPT-User in your server logs by this string. At the checked-on date OpenAI publishes: Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; ChatGPT-User/1.0; +https://openai.com/bot. The token is ChatGPT-User.
Does ChatGPT-User train OpenAI's models?
OpenAI does not document a training use for ChatGPT-User, and Prefer's free AI Crawler Access Checker lists GPTBot as the OpenAI training crawler. OpenAI's training opt-out is the GPTBot token.
Read the GPTBot entry →How do I know if ChatGPT-User visited my site?
Prefer's Agent Analytics shows ChatGPT-User visits from your server logs. To check by hand, search your access logs for the ChatGPT-User token, then confirm the IP falls inside the ranges at openai.com/chatgpt-user.json.
Questions
Asked plainly.
What is ChatGPT-User?
ChatGPT-User is the user agent OpenAI uses for certain user actions in ChatGPT and Custom GPTs, and Prefer's Agent Analytics shows its visits in your server logs. When a user asks ChatGPT or a Custom GPT a question, it may visit a web page with the ChatGPT-User agent. OpenAI says it is not used for crawling the web in an automatic fashion.
Does ChatGPT-User respect robots.txt?
Not reliably, and Prefer's free AI Crawler Access Checker flags this when it reads your robots.txt. OpenAI says that because these actions are initiated by a user, robots.txt rules may not apply. A Disallow rule for ChatGPT-User is a signal, not a guaranteed block.
Does blocking ChatGPT-User remove me from ChatGPT search?
No, and Prefer's free AI Crawler Access Checker shows which OpenAI token actually controls search. OpenAI says ChatGPT-User is not used to determine whether content may appear in search, and points site owners to OAI-SearchBot in robots.txt for search opt-outs and automatic crawl.
Should I allow ChatGPT-User?
For most brands, yes, and Prefer's free robots.txt generator can write the rule for you. When a ChatGPT user asks about your page, ChatGPT-User is how ChatGPT reads it, so blocking it can leave that person without your content. OpenAI does not document any training use for ChatGPT-User.
Sources
Where these facts come from.
Every fact on this page traces to one of these. Where OpenAI does not document something, the page says so.
- 01 OpenAI, overview of OpenAI crawlers The ChatGPT-User user agent string and purpose, the note that robots.txt rules may not apply, and the pointer to OAI-SearchBot for search opt-outs. Vendor docs 1 Oct 2026
- 02 OpenAI, ChatGPT-User IP ranges The published IP list used for the verification step above. Vendor data 1 Oct 2026
- 03 Robots Exclusion Protocol, RFC 9309 The standard that defines how user-agent tokens, Allow and Disallow rules are interpreted. Standard 1 Oct 2026
ChatGPT and OpenAI are trademarks of OpenAI. Prefer is not affiliated with, endorsed by or sponsored by OpenAI. This entry reflects public documentation at the dates shown. Something wrong here? Tell us and we will fix it →
Keep reading