Photo: Getty Images
More than a third of English-language web pages published since ChatGPT launched in November 2022 show signs that they were written or substantially edited by artificial intelligence, according to a new Pew Research Center analysis published August 20.
The study used the Common Crawl web archive and examined nearly 500,000 English-language pages published over roughly five years, including material from before and after ChatGPT’s launch.
Researchers used OpenAI Pangram’s detection technology to estimate whether pages were likely written or significantly edited by AI.
In a random sample of 10,000 web pages collected in July 2026, around 10% showed significant signs of AI authorship. However, that sample also included pages published before generative AI writing tools became widely available.
After filtering out older pages and focusing only on content published after ChatGPT’s launch, the proportion rose to 35%.
AI use varies by domain
The analysis found significant differences between website domains.
Signs of AI authorship were detected around 10 times more often on .com sites than on .edu and .gov sites, where the rate was about 1%. For .org domains, the figure was 4.6%.
Pew researchers caution that AI detectors are not perfectly reliable. Pangram, like other detection systems, can incorrectly classify human-written material as AI-generated.
Still, researchers said the findings point to a broader trend: AI is rapidly becoming a major source of new content across the internet.
The results follow other studies suggesting that AI-generated writing has already become widespread online. A 2025 Graphite study found that more than half of written online content was generated by AI, while separate Pangram research found substantial AI-generated content across social media.
