GPTBot
OpenAI · Collects for training
OpenAI's setting for GPTBot is independent of OAI-SearchBot's: you can keep your content out of model training and still appear in ChatGPT search. If you allow both, OpenAI may use a single crawl for both purposes.
Fact sheet
- Operator
- OpenAI
- What it is for
- It collects text that is used to train AI models.
- Assistant it feeds
- Does not feed any assistant's answers
- Counts towards the Parsigo score
- No. What it collects is used to train models, and training does not decide whether you are cited.
- Official documentation
- developers.openai.com/api/docs/bots
- Official IP ranges
- JSON list
If you block it
You ask it to stop collecting your content. It changes neither whether you are cited nor your score.
Allow or block
A group of its own makes GPTBot ignore the “User-agent: *” rules. If you have Disallow lines there that should still apply to it, repeat them inside its group. To allow it, writing nothing is usually enough.
Allow
User-agent: GPTBot Allow: /
Block
User-agent: GPTBot Disallow: /
Test it on your site
We run the diagnostic on your site and show you whether GPTBot can get in, with the robots.txt line that decides it. Free, no sign-up.
Worth knowing
Collects data to train models. Blocking it is a legitimate editorial decision and does not stop ChatGPT citing your site.