GPTBot

What user agent does GPTBot use?

OpenAI publishes this user agent string for GPTBot.

OpenAI says the version number may change.

Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; GPTBot/1.4; +https://openai.com/gptbot

How do I block GPTBot?

OpenAI says disallowing GPTBot tells it your content should not be used to train its models. It does not take your site out of ChatGPT search, which OAI-SearchBot controls separately.

Add this to the robots.txt file at the root of your site:

User-agent: GPTBot
Disallow: /

How do I check that a GPTBot visit is real?

Anyone can put a crawler’s name in a user agent, so the name alone proves nothing.

OpenAI publishes the IP addresses GPTBot crawls from, as JSON.

Spreelytics runs this check for you. Every crawl that claims to be GPTBot is compared with the IP ranges OpenAI publishes, refreshed every 12 hours, and one from outside them counts as spoofed.

How do I see GPTBot visits on my site?

Your server’s access logs record the user agent of each request, so you can search them for “GPTBot”.

Or add the Spreelytics server middleware, @spreelytics/ai-crawl. It recognises GPTBot by its user agent and counts each visit on the Bots page under Training, apart from your visitor numbers.

Sources: developers.openai.com/api/docs/bots