What GPTBot does
GPTBot crawls public web pages to collect training data for OpenAI's models. It is not the agent that answers ChatGPT users in real time — that's ChatGPT-User — and it is not ChatGPT search's index crawler either (that's OAI-SearchBot). It fetches raw HTML and does not execute JavaScript.
Fact sheet
- Operator
- OpenAI
- robots.txt token
- GPTBot
- Honors robots.txt
- Yes (documented)
- Executes JavaScript
- No — invisible to browser analytics
- Identity verifiable by IP
- Yes — vendor publishes IP ranges (we verify every visit)
- Official documentation
- OpenAI docs ↗
How it identifies itself
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; GPTBot/1.2; +https://openai.com/gptbot
User-agent strings are free to fake — any scraper can claim to be GPTBot. That's why VisitorType checks every claimed GPTBot visit against OpenAI's published IP ranges and downgrades impostors to plain bots.
Allow or block via robots.txt
# Block GPTBot User-agent: GPTBot Disallow: / # Or allow it explicitly User-agent: GPTBot Allow: /
Should you block GPTBot?
Blocking GPTBot only opts your content out of model training — you stay visible in ChatGPT search and reachable by users' assistants. It's a licensing/philosophy decision, not a traffic decision. Measure how often it reads you and which pages before deciding.
Is GPTBot reading your site?
GPTBotnever runs JavaScript, so GA4 can't see it. VisitorType's server-side detection can — with per-page counts and IP-verified identity.
Measure it free