visibian.HQ
visibian.com

OpenAIChecked 29 Oct 2026

GPTBot

Collects pages that may be used to train OpenAI's AI models.

Run by
OpenAI
Used for
Training AI
Follows robots.txt
Yes
Shut out by
13% of top websites

What blocking it changes

OpenAI reads it as a request not to use the site's pages to train its models. It does not keep a site out of ChatGPT search: that is OAI-SearchBot.

"Disallowing GPTBot indicates a site's content should not be used in training generative AI foundation models."

OpenAI, read 29 Oct 2026
Figure 1How many shut out GPTBot while letting Googlebot in
Most visited websites13%89 of 669
S&P 500 companies4%15 of 410
From our two studies of 29 October 2026: the 669 most visited websites and the 410 S&P 500 companies whose robots.txt we could read. 131 of the websites and 41 of the companies name GPTBot in the file. The website study, the S&P 500 study.

The lines to copy

To shut it out
User-agent: GPTBot
Disallow: /
To let it in
User-agent: GPTBot
Allow: /

A crawler follows only the lines under the User-agent line that names it, and the lines under User-agent: * only when no line names it. So lines for one crawler do not change the rules for the others.

OpenAI publishes the addresses it crawls from, so a request can be checked against them: the list.

More from OpenAI

  • OAI-SearchBot Finds the pages that ChatGPT search can show and link to.
  • ChatGPT-User Opens a page when someone asks ChatGPT, or a custom GPT, about it.