AI & Models
Pew: over a third of post-ChatGPT web pages show signs of AI authorship
A Pew Research study found that more than a third of English-language web pages published since ChatGPT's launch show signs of being written or heavily edited by AI.
The report, released Thursday, corroborates other studies tracking how much of the web’s newer content is now AI-generated or AI-edited. It arrives shortly after internet infrastructure provider Cloudflare reported that bot web traffic had overtaken human web traffic — a milestone Cloudflare said it reached sooner than expected. Where Cloudflare measured who is browsing, Pew measured what is being browsed, and found much of it is bots reading pages written by other bots.
To compile the report, Pew used the Common Crawl web archive to collect nearly half a million English-language web pages spanning roughly the past five years, starting a couple of years before ChatGPT’s November 2022 release, then used Open Pangram’s detection technology to flag pages likely written or heavily edited by AI. In a random sample of 10,000 pages collected in July 2026, around 10% showed significant signs of AI authorship — but Pew noted that a random sample like this inevitably pulls in older pages published before AI writing tools even existed.
Filtering out those pre-ChatGPT pages and looking only at pages published after its release, Pew found signs of AI authorship in over one-third (35%) of the sample. The rate varied sharply by domain type: .com pages showed signs of AI authorship at around 10 times the rate of .edu or .gov pages, both of which came in near 1%. .org domains sat in between, at a 4.6% AI-authorship rate.
Pew cautioned that AI-detection tools like Pangram can misclassify human-written pages as AI-generated, so the figures aren’t precise — but said the data is likely at least directionally correct at this scale. The study also found that other purported AI-writing tells, including em dashes, Oxford commas, and phrasing like “it’s not X, it’s Y,” have become more common across the years studied.
Why it matters
A web where over a third of new pages carry AI’s fingerprints complicates everything downstream that assumes web text reflects human judgment — from search rankings to the training data for the next generation of AI models.