In a random sample of 10,000 webpages collected in July 2026, one in ten shows signs of being written or substantially edited by AI, according to a new analysis from the Pew Research Center published on August 20. When researchers narrowed the sample to content published after ChatGPT’s November 2022 launch, the share jumped sharply: signs of AI authorship appeared in more than one-third, or 35%, of that more recent content.
Commercial sites are where the concentration is highest
The concentration of AI-written content is not spread evenly across the web, and the gap carries direct implications for everyday consumers. When ChatGPT was first released, the kinds of linguistic patterns that can signal AI authorship appeared at similar rates across the main top-level web domains.
In samples from 2026, around one-in-ten pages with a .com domain show signs of AI authorship, about double the share on .org domains (4.6%), and ten times the rate on .edu or .gov domains (both around 1%).
Commercial sites more frequently publish large volumes of text that can be relatively easily automated: product descriptions, reference articles, SEO content, and news. That concentration means shoppers browsing product pages, reading buyer guides, or comparing reviews online are far more likely to be reading machine-generated prose than users visiting academic or government sites.
How Pew measured the shift
The analysis drew on Pew Research Center’s examination of 490,000 English-language webpage texts sampled from the Common Crawl web repository and analyzed using Open Pangram. The data covers Common Crawl web archive snapshots collected between January 2021 and July 2026.
To analyze the data, researchers used an open-weight AI detection model created by Pangram, which looks for statistical differences in how human authors and AI models use language, helping identify when a text shows signs of AI authorship.
Pew was candid about the method’s limits: detectors get individual documents wrong, and Pew says so, treating its argument as statistical, across hundreds of thousands of pages the patterns separate reliably even when any single verdict might not.
The findings are in line with other studies that have shown that large shares of recently published pages on the internet were likely written or substantially edited by AI. That broader comparison is derived by a different methodology but aligns with Pew’s own finding that 35% of pages in the July 2026 crawl with post-ChatGPT publication dates show signs of AI authorship.
Stylistic tells are accumulating
Beyond raw detection scores, Pew’s team tracked specific writing patterns that have trended upward alongside AI adoption. Compared with 2023, em-dash usage frequency roughly doubled, while Oxford comma usage increased by 63%. Formal vocabulary preferred by AI, such as “delve,” “interplay,” and “testament”, more than doubled in frequency, and contrastive sentence structures of the “not simply X, but Y” variety nearly tripled in usage.
Those patterns are not conclusive on their own, but at the scale of hundreds of thousands of pages the directional signal tracks consistently with the authorship data.
Of all the pages in the sample, 10% show significant signs of AI authorship, part of an upward trend that began in late 2022 when ChatGPT was first released and continued as other AI chatbots such as Claude and Gemini followed. The Pew findings also arrive shortly after internet infrastructure provider Cloudflare reported that bot web traffic had overtaken human web traffic, a milestone that was reached sooner than the company had estimated.
What it means for consumers browsing online
The practical effect for shoppers and everyday web users is a gradual erosion of the assumption that a person wrote what they are reading. As generative AI tools continue to become more sophisticated and more deeply embedded in everyday content production workflows, researchers, publishers, and advertisers alike are likely to continue tracking how the balance between human and AI-assisted authorship evolves.
That question is particularly pressing given how dramatically the share of AI-influenced content has grown in roughly three and a half years since ChatGPT’s initial release.
Consumers who rely on the open web to research purchases, compare prices, or read product reviews are increasingly navigating a landscape where the prose they encounter may have been generated with little or no human editorial judgment applied afterward.
The Pew analysis is a useful reminder that online shopping behavior is shifting in more ways than cart abandonment rates capture, and that the information environment shaping those decisions is itself changing in fundamental ways.
The study’s domain-level breakdown suggests that the users most exposed to AI-written content are those visiting everyday commercial websites. People relying on constant internet access for research or social connection encounter an increasingly machine-authored web with each passing month.
The full Pew Research Center report, including methodology, is available at pewresearch.org.
