A new Pew Research study finds that over one-third of web pages published since ChatGPT’s release show significant signs of AI authorship. The report corroborates earlier studies showing that much newer web content is either written or heavily edited by AI, and it arrives shortly after Cloudflare reported that bot traffic has overtaken human web traffic. Pew’s look is not at who is browsing, but at what is being published.

Pew built its findings using the Common Crawl archive, gathering nearly 500,000 English-language web pages spanning roughly the past five years, starting a couple of years before ChatGPT’s November 2022 release. The pages were run through AI-author detector based on Pangram‘s technology. In a random sample of 10,000 pages collected in July 2026, about 10% showed significant signs of AI authorship. That sample necessarily included older pages that predate AI writing tools, so Pew also looked only at pages published after ChatGPT’s release; in that filtered set the share rose to 35%.

Domain differences were stark. URLs ending in .com showed signs of AI authorship at roughly 10 times the rate of .edu or .gov domains, both around 1%, while .org domains had a 4.6% rate. Pew also notes that no automated detector is perfect: tools can misclassify human-written content as AI-generated, so the figures should be treated as directionally correct at scale rather than exact per page. The report also found that common stylistic ties of AI writing have increased over time, including em dashes, Oxford commas, and formulations like “it’s not X, it’s Y.”