AnswerIndex

Original measurement data on how the web treats AI crawlers.

AI crawler data · November 2026

GPTBot

Operator: OpenAI · Purpose: Training & data collection · Measured November 19, 2026 by AnswerIndex

GPTBot is OpenAI’s crawler. It collects web content that may be used to train and improve OpenAI’s models.

In our November 19, 2026 measurement of the Tranco top 10,000 websites, 1,764 of the 7,091 sites with a determinable robots policy block GPTBot (24.9%), while 5,327 allow it (75.1%). That ranks GPTBot #3 of the 14 AI agents measured, by blocked share.

24.9%
of sites with a determinable robots policy block GPTBot
1,764
sites block GPTBot (of 7,091 determinable)
5,327
sites allow GPTBot (75.1%)
#3
of 14 AI agents, ranked by blocked share

GPTBot in context

How to block or allow GPTBot in robots.txt

robots.txt rules are matched by user-agent token. To block GPTBot from an entire site:

User-agent: GPTBot
Disallow: /

To explicitly allow it (the default when no group mentions the token):

User-agent: GPTBot
Allow: /

AnswerIndex evaluates these rules per RFC 9309: an agent-specific group beats the * group, and a site whose robots.txt refuses with HTTP 401/403 counts as fully disallowed for every agent. Note that robots.txt is advisory — it records the policy a site declares, and user-action agents fetching at a person’s request may not be bound by it.

Source data

These figures come from the AI Crawler & llms.txt Adoption Index — November 2026, based on the Tranco top 10,000 measured on November 19, 2026. Compare all agents on the AI crawlers index, or download the full per-domain dataset: results-2026-11-19.csv.

Cite this: AnswerIndex, AI Crawler & llms.txt Adoption Index, November 2026. https://answerindex.org/reports/2026-11-ai-crawler-index/