OpenAI crawlers
OpenAI operates 3 crawlers that may reach your site. They do different jobs, so a single blanket rule for all of them is almost always the wrong call.
| User-agent | Type | What it does |
|---|---|---|
GPTBot | Training | Collects pages to train future models. |
OAI-SearchBot | Search | Builds the index an assistant searches when it answers. |
ChatGPT-User | User-triggered | Fetches your page because someone asked about it just now. |
What blocking each one costs
- Training
- Blocking costs nothing today, but means the next model has never heard of you.
- Search
- Blocking removes you from answers happening right now.
- User-triggered
- Blocking means that person gets told the page could not be read.
robots.txt
To let every OpenAI crawler through:
User-agent: GPTBot
Allow: /
User-agent: OAI-SearchBot
Allow: /
User-agent: ChatGPT-User
Allow: /To block all of them:
User-agent: GPTBot
Disallow: /
User-agent: OAI-SearchBot
Disallow: /
User-agent: ChatGPT-User
Disallow: /robots.txt is honoured voluntarily. It is the right way to state your intent, and it is not an access control — treat anything you genuinely need kept out as needing authentication.
Sources
Is OpenAI allowed on your site?
We'll read your robots.txt and tell you which AI crawlers it lets through. No signup.