Tallyhouse

How many of the top 1000 websites tell AI crawlers to stay out?

Three years ago almost no major website told AI crawlers to stay out. By 2026-08, more than a quarter did — and the change was not gradual.

A fixed panel of 611 long-running popular sites, read from Common Crawl's archive, stood at 1.28% in 2023-01. The largest single move in the series is 2023-09, up 13.65 points on the reading before it — the first snapshot taken after OpenAI launched GPTBot in August 2023. The figure reached 27.85% by 2026-08.

Trend

Targeted and effective access over time, 16 quarterly periods from 2023-01 to 2026-08.Targeted access rose from 1.28% at 2023-01 to 27.85% at 2026-08. Effective access rose from 3.08% at 2023-01 to 29.71% at 2026-08.0%10%20%30%2023-012023-092024-052025-012025-102026-0827.85%29.71%

The weekly Agent Accessibility Index currently reads 24.32% for the week of 2026-10-05.

Read Three years of AI-crawler blocking for the full three-year series and the method behind it.

Note: a different population from the live index panel, so levels are not comparable between the two indices.

How it is built and why the numbers can be trusted