The Pew Research Center analyzed nearly half a million English-language web pages for AI-generated content. Since ChatGPT launched, the share of machine-written text online has climbed sharply.

The texts came from the Common Crawl web archive and were checked for signs of machine authorship using the detection tool Open Pangram. In a sample from July 2026, about 10 percent of all pages examined showed clear signs of AI authorship.

Filtering the sample to only pages published after ChatGPT's release changes the picture dramatically: more than a third of those newer pages show signs of AI authorship. The trend kicked off with ChatGPT in late 2022, and the share of likely AI-generated web content has climbed steadily ever since.

A sharp split by domain

The distribution gives a clear signal about where this content is produced:

  • Commercial domains — about one page in ten shows signs of AI authorship.
  • Organization domains — 4.6 percent.
  • School and government domains — roughly 1 percent each.

That makes commercial websites roughly ten times more likely to contain AI-written text than pages from schools or government agencies. The gap says less about where the tool is useful than about where the commercial pressure is: wherever content production is tied directly to revenue, volume outruns sourcing.

The style is changing too

Pew's analysis found several language patterns that have become much more common on the web since 2023. Em dashes now show up about twice as often as they did in 2023, and Oxford comma usage has jumped 63 percent.

Certain AI-favorite words — "delve," "interplay," "testament," "pivotal," "landscape," "tapestry," "bolstered," "crucial," "meticulous," "vibrant" — have more than doubled in frequency. Negative parallelisms following the "it's not just X, it's Y" pattern have nearly tripled, though they remain rare in absolute numbers. A separate study of corporate PR documents found that phrase quadrupled since 2022.

A study by Imperial College London, the Internet Archive and Stanford University from April 2026 reached a similar conclusion, finding roughly 35 percent of newly published websites were fully or partly AI-generated. The researchers also found 33 percent higher semantic similarity between AI texts and a much more positive tone overall, but cautioned that public perception of negative effects often goes well beyond what the data supports.

What counts as "AI text"

There is a problem with this and similar studies, and with the public debate: nobody agrees on what "AI text" even means. The spectrum runs from fully automated content to human drafts polished with AI to texts where a model stepped in for a few sentences.

Open Pangram and similar models can only make a rough call on whether a human or a machine likely wrote something. They cannot reliably say how much AI was involved or at what stage, and they still misfire regularly. Yet these are very different ways of working.

Why it matters

Public debate around AI text is growing more polarized. Using the tools already carries a stigma that cuts both ways: those who disclose pay a price, and so do those who conceal. Neither camp leaves much room for the messy reality of how people actually write with these tools.

And since adoption is not slowing down, finding that middle ground is going to matter more and more. This site is itself one version of that middle ground: the texts are written with AI, the sources are read, the facts are checked, and a notice saying so sits under every article.