Pew Study Finds AI-Written Text Has Surged Across the Web Since ChatGPT

The Pew Research Center analyzed nearly half a million English-language web pages for signs of AI-generated content. The texts came from the Common Crawl web archive and were checked with the Open Pangram detection tool. In a July 2026 sample, about 10 percent of all pages examined showed clear signs of AI authorship. Among pages published after ChatGPT’s release, more than a third showed signs of AI authorship.

The analysis found a major gap by domain type. About one in ten .com pages showed signs of AI authorship, compared with 4.6 percent of .org pages and about 1 percent each for .edu and .gov pages. That makes commercial websites roughly ten times more likely to contain AI-written text than school or government sites. For readers, the takeaway is simple: domain context matters when judging how likely a page is to include machine-written material.

Pew also found that certain words, phrases and punctuation patterns have become more common on the web since 2023. Em dashes now appear about twice as often as they did in 2023, while Oxford comma usage has jumped 63 percent. Words such as “delve,” “interplay,” “testament,” “pivotal,” “landscape,” “tapestry,” “bolstered,” “crucial,” “meticulous” and “vibrant” have more than doubled in frequency. The “it’s not just X, it’s Y” pattern has nearly tripled, though it remains rare in absolute terms.

The article stresses that “AI text” is still a fuzzy category. It can mean fully automated writing, human drafts polished with AI, or text where a model helped with only a few sentences. Open Pangram and similar tools can make rough calls about likely machine authorship, but they cannot reliably show how much AI was involved or at what stage. As AI adoption grows, the practical challenge is separating useful transparency from overconfident labeling.