Knocklog

GPTBot

OpenAI's AI training crawler.

Operator
OpenAI
Type
AI training crawler
In the user agent
GPTBot
robots.txt token
GPTBot
Follows robots.txt
says it follows robots.txt
Official documentation
developers.openai.com
Checked on
2026-11-04

What is GPTBot?

GPTBot is OpenAI's AI training crawler. It collects pages that may be used to train AI models. A visit does not answer anyone right now; it decides what future models know about you.

The operator's description: Crawls content that may be used to train OpenAI's generative AI foundation models. (developers.openai.com)

How do I block or allow GPTBot?

With robots.txt, using the token GPTBot. OpenAI says it follows robots.txt. To block the whole site:

User-agent: GPTBot
Disallow: /

To allow it, use Allow instead of Disallow. Before blocking, weigh what it brings: blocking does not remove you from AI search, but future models will know less about you.

How do I see whether GPTBot visits my site?

In your server or CDN access logs, look for GPTBot in the user agent. Google Analytics does not show these visits: agents do not run its JavaScript.

Anyone can copy the name in a user agent, so counting log lines is not enough. Knocklog reads your site's traffic on any host, confirms who is who, and shows how often GPTBot came, which pages it read and what that brought you.

How do I know it is really GPTBot?

OpenAI publishes GPTBot's IP addresses at https://openai.com/gptbot.json. A visit that claims to be GPTBot from outside those ranges is spoofed.

Knocklog runs this check on every visit and separates the genuine agent from impostors.

Does GPTBot visit your site?

See in seconds how your site treats AI agents. Then connect your traffic to learn who really came.

Other OpenAI agents

All AI agents