Botservatory

← Catalogue

OpenAI

GPTBot

Class
Training
robots.txt
Says it honours robots.txt
Disallowing GPTBot opts your content out of training.
Evidence
Confirmed from operator documentation
Source
Published User-Agent
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; GPTBot/1.4; +https://openai.com/gptbot

Proving it is real

Check an address now.

Blocking it

robots.txt
User-agent: GPTBot
Disallow: /
Cloudflare custom rule expression
(http.user_agent contains "GPTBot")
nginx
if ($http_user_agent ~* "GPTBot") {
    return 403;
}

Matching a name only blocks bots that tell the truth about it. Before you do this, read what each decision costs.

Last reviewed Nov 2, 2026