About a quarter of crawlable webpages now contain AI-generated text, study estimates

Researchers from the University of Washington, UC Berkeley and other universities posted a study to arXiv estimating that in August 2026 about 24% of crawlable webpages (95% credible interval 22.9–24.4%) contained text fully generated or edited by AI, up from near zero before 2022. The estimate rests on a sample of 5.1 million pages from 71 Common Crawl snapshots between January 2019 and August 2026, scored with AI detectors and a statistical model that corrects for the detectors’ false positives and false negatives. The estimated share was 32.8% of English-only pages against 17.0% of pages with no English, it was lowest on the web’s most linked-to domains, and the two least-linked tiers, holding 70.5% of pages, accounted for 78.8% of the increase.