Study finds 35% of newer webpages show signs of AI writing

In short: A Pew Research study estimates that about 35% of English webpages published after ChatGPT’s launch show signs of AI writing or heavy AI editing.

What's going on

Pew Research looked at how much of today’s web content appears to be written by AI tools like ChatGPT. The group analyzed nearly half a million English-language webpages from the past five or so years.

To do this, Pew pulled pages from Common Crawl, a public web archive (think of it like a giant library snapshot of the internet). Then it used detection technology from Open Pangram to flag pages that were likely written or “substantially edited” by AI.

In a random sample of 10,000 webpages collected in July 2026, about 10% showed “significant signs of AI authorship.” But Pew said that number is lower because a truly random sample includes older pages, including ones from before ChatGPT launched in November 2022.

When Pew filtered the data to focus only on pages published after ChatGPT’s release, the share with signs of AI authorship rose to 35%. Pew also found big differences by domain name. Pages on .com sites showed signs of AI authorship about 10 times more often than pages on .edu or .gov sites, which were both around 1%. .org sites were at 4.6%.

What to watch

AI detection tools can be wrong, so these numbers are not exact. Still, the study suggests more of what people read online may be written by software, not people. It also raises a practical question for readers and publishers, which is how to judge quality when bots may be writing pages for other bots to read.

Source: TechCrunch AI