A Pew study finds AI-written web content has risen sharply since ChatGPT's launch.
Over a third of newer web pages show signs of AI authorship.
Commercial .com sites are roughly ten times more likely to contain AI text than educational or government sites.
What happened
Pew Research Center analyzed nearly half a million English-language web pages and found that since ChatGPT's launch, the share of machine-written text has climbed sharply. In a July 2026 sample, about 10% of all pages showed clear signs of AI authorship, while more than a third of pages published after ChatGPT's release did.
Why it matters
Commercial .com sites are about ten times more likely to contain AI-written text than .edu or .gov pages, which sit at about 1% each. Common AI-favorite words like "delve," "pivotal," and "landscape" have more than doubled in frequency, and em dash usage has roughly doubled since 2023.
What to watch
A separate April 2026 study by Imperial College London, the Internet Archive, and Stanford University found that roughly 35% of newly published websites were fully or partly AI-generated. It also found 33% higher semantic similarity between AI texts and a much more positive tone overall, but cautioned that public perception of negative effects often exceeds what the data supports.
Ask the AI about this article →
The Pew Research Center's analysis of nearly half a million English-language web pages from the Common Crawl archive provides concrete data on the spread of AI-generated text. The study's key finding is the dramatic difference between the overall rate of AI authorship (about 10 percent) and the rate for pages published after ChatGPT's release in late 2022 (more than a third). This points to ChatGPT's launch as a clear inflection point for machine-written content online.
The research also quantifies language patterns associated with AI text, noting that words like "delve," "interplay," and "pivotal" have more than doubled in frequency since 2023. However, the article highlights a significant caveat: there is no consensus on what "AI text" actually means, and detection tools like Open Pangram can only make a rough call on authorship. This fuzziness complicates the public debate, which the article notes is growing more polarized, as seen in the discussion around Anthropic's planned watermark for Claude output. The reality of how people write with AI tools likely falls on a spectrum, from fully automated content to human drafts that are merely polished, a nuance that both camps in the debate often miss.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
AWS last month rolled out cloud-to-edge solutions for customers building physical AI systems—AI that perceives…
Hugging Face, backed by Nvidia, is reportedly seeking a deal that could value it at $13 billion or more, and h…
Broadcom is reportedly in talks to raise between $70 billion and $80 billion in debt, a package that could eve…
Adobe recorded a 1,200% increase in traffic to U.S
OpenAI announced in mid-August that Dali Rajic, the president and COO at Google-owned cybersecurity company Wi…
Instinct, an AI personal assistant in private testing, has drawn praise for its capabilities but also criticis…
