A third of web pages published since ChatGPT’s launch show signs of AI authorship, study finds
SkimNews Take
AI authorship clusters on .com domains likely because institutional sites (.gov, .edu) carry reputational and editorial accountability, while commercial publishers face incentives favoring cheap volume — suggesting AI-generated content fills precisely the spaces where verification is weakest.
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Pew Research found that over one-third (35%) of web pages published after ChatGPT's November 2022 release show signs of AI authorship, based on analysis of nearly half a million English-language pages from the Common Crawl web archive.
- A random sample of 10,000 web pages collected in July 2026 showed only about 10% with "significant signs of AI authorship" because it included older pages from before AI writing tools existed.
- .com domains showed AI authorship signs at roughly 10x the rate of .edu and .gov domains (around 10% vs approximately 1%), while .org domains came in at 4.6%.
- Pangram's AI-detection technology identified the patterns, though Pew acknowledged the tool can misclassify pages and said the results are only "likely at least directionally correct" at scale.
- Stylistic AI tells like em dashes, Oxford commas, and "it's not X, it's Y" phrasing also rose over the years, according to Pew.
- The study arrives shortly after Cloudflare reported that bot web traffic had overtaken human web traffic — a milestone the company said was reached sooner than it had estimated.
Why it matters: The roughly 10x gap between .com pages and .edu/.gov domains suggests commercial publishers are flooding the web with AI-generated text while institutional sites largely haven't — meaning the average reader landing on a commercial URL is far more likely to encounter bot-written content than when browsing government or university resources.
Ask SkimNews
