AI SearchOpenAI
Last verified: 2026-07-16 · maintained by Unsourced
OAI-SearchBot powers ChatGPT Search results. It fetches pages to surface and link them in live answers, so it can drive real referral traffic — one to court, not block.
OAI-SearchBot is the indexing crawler behind ChatGPT's search feature — it builds the pool of pages ChatGPT can pull from and link when it answers with live results. Unlike GPTBot it isn't gathering training data; its job is retrieval, so a page it has indexed can appear as a cited, clickable source. Blocking it forfeits that placement while doing nothing either way about model training.
OAI-SearchBot can be checked two independent ways, and a real request passes both. Its source IP has to fall inside the 35 ranges OpenAI publishes, and it has to forward-confirm by reverse DNS: the IP resolves to openai.com, and that hostname resolves back to the same IP. Anything carrying the OAI-SearchBot user-agent that fails either test is an impostor, whatever the header says.
Recommended: bait. It's the retrieval index behind ChatGPT Search, so allowing it keeps your pages eligible as linked, clickable sources; a block forfeits that placement and changes nothing about model training either way.
If you do choose to act in robots.txt (which crawlers honour but don't enforce):
# OAI-SearchBot: recommended to ALLOW — blocking can cost you AI visibility User-agent: OAI-SearchBot Disallow:
What does OAI-SearchBot actually do?
It builds the index behind ChatGPT Search, so the pages it collects are the ones ChatGPT can surface and link in live results. Being indexed by it is what makes you eligible to be cited.
Does blocking OAI-SearchBot affect model training?
No — it's purely a retrieval crawler, separate from GPTBot's training collection. Blocking it forfeits your place in ChatGPT Search while doing nothing either way about what models learn.
How do I confirm a request is really OAI-SearchBot?
Check the source IP against OpenAI's published ranges and run a reverse-DNS lookup to openai.com — a request that matches neither is wearing the name, not using it.
Unsourced audits the requests reaching your site, separates genuine OpenAI traffic from spoofed headers, and tracks whether AI answers cite you.
Check your site free →10-day free trial · no card required · cancel anytime