A third of webpages published since ChatGPT’s launch show signs of AI authorship, study finds
TechCrunch Sarah Perez
Pew says over a third of newer webpages show AI writing signs. That means the web is now full of bots writing for bots.
Based on reporting by TechCrunch, Sarah Perez — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Pew Research says a big chunk of the newer web looks machine-made. In a study released Thursday, it found that more than one-third of webpages published after ChatGPT launched show signs of AI authorship. The headline is stark, but the method matters too: Pew is looking at what’s on the page, not who’s visiting it.
To build the report, Pew pulled nearly half a million English-language webpages from Common Crawl, reaching back to a couple of years before ChatGPT’s November 2022 debut. It then used Open Pangram’s detection tech to judge whether pages were likely written or heavily edited by AI. In a random sample of 10,000 pages collected in July 2026, about 10% showed significant signs of AI authorship. Pew also said that sample included older pages, which naturally couldn’t have been written by tools that didn’t exist yet.
So Pew narrowed the lens. Once those older pages were removed, the share with signs of AI authorship rose to 35%. That lines up with other recent work suggesting newer web pages are increasingly being written, or at least heavily polished, by AI. And it lands right after Cloudflare said bot traffic had passed human traffic on the internet, earlier than it had expected.
The study also found a sharp split by domain. Pages on .com showed signs of AI authorship at roughly 10 times the rate of .edu or .gov sites, which were both around 1%. .org sites came in at 4.6%. Pew says the tool can misclassify pages, but at this scale the direction of travel is hard to ignore. Even the writing tells on itself: more em dashes, more Oxford commas, more of that tired “it’s not X, it’s Y” phrasing. The robots are learning the house style.
My take — AI-written commentary, not fact-checked reporting
This is what happens when everyone uses the same machine to sound smart: the internet starts to read like a committee memo with better grammar. The real joke is that .com is carrying the AI slop burden while .edu and .gov still look like they remember what a human typo is. The web didn’t just get noisier; it got self-referential in the laziest possible way.
Read more about this at: TechCrunch
