Study: 35% of post-ChatGPT web pages show AI authorship signs
Original: A third of web pages published since ChatGPT’s launch show signs of AI authorship, study finds
Why This Matters
The findings quantify AI's rapid displacement of human-authored content across the open web, with broad implications for information quality and AI training data integrity.
Pew Research released a study on August 20, 2026, finding that over one-third (35%) of English-language web pages published after ChatGPT's November 2022 launch show significant signs of AI authorship, based on analysis of nearly 500,000 pages from the Common Crawl archive.
Pew Research used the Common Crawl web archive to collect nearly 500,000 English-language web pages spanning roughly five years, starting before ChatGPT's November 2022 release. Using Open Pangram's AI-detection technology, Pew found that in a random sample of 10,000 pages captured in July 2026, about 10% showed significant signs of AI authorship. When older, pre-ChatGPT pages were filtered out, that figure rose to 35%. Domain-level analysis revealed stark differences: .com domains showed AI authorship at roughly 10 times the rate of .edu and .gov domains (each around 1%), while .org domains sat at 4.6%. Pew also noted a rise in stylistic markers associated with AI writing, including increased use of em dashes, Oxford commas, and constructions like 'it's not X, it's Y.' The study arrives shortly after Cloudflare reported that bot web traffic had surpassed human web traffic — a milestone reached earlier than the company had projected. Pew acknowledged that AI-detection tools like Pangram can produce false positives, but stated the findings are likely directionally accurate at scale.