A new analysis from Pew Research Center suggests that artificial intelligence has quietly become a co-author of the modern internet, with roughly a third of webpages created since the launch of ChatGPT bearing the hallmarks of machine-generated text.
What Pew Discovered
Pew researchers ran nearly half a million webpages through an AI-detection tool to gauge how deeply automated writing has seeped into online content. The study focused on pages published after ChatGPT's debut in late 2022, a moment widely regarded as the tipping point for mass-market generative AI.
The findings point to a rapid and accelerating shift. A substantial share of newly created pages carried signatures consistent with AI authorship, and that proportion has been climbing steadily rather than leveling off.
Roughly one in three pages born after ChatGPT now shows the fingerprints of a machine.
The concentration was especially pronounced on commercial domains. Pages ending in .com were among the most likely to contain AI-generated material, a pattern that hints at the commercial incentives driving the flood of automated content across the web.
Why It Matters
The results underscore how quickly generative tools have reshaped the economics of publishing online. Producing text at scale has never been cheaper, and businesses appear to be leaning heavily on AI to churn out marketing copy, product descriptions, and filler content.
That raises thorny questions for readers, search engines, and the broader information ecosystem. As machine-written pages proliferate, distinguishing authentic human expertise from automated output becomes harder—and the risk of low-quality or misleading content spreading grows.
A few takeaways stand out from the research:
- AI-written content is concentrated on commercial .com domains.
- The share of machine-generated pages is rising quickly, not plateauing.
- Detection tools remain imperfect, so figures should be read as estimates.
Pew's analysis relies on AI-detection technology, which is not infallible and can produce both false positives and false negatives. Even so, the study offers one of the clearest snapshots yet of just how thoroughly generative AI has embedded itself into the fabric of the web in a remarkably short span of time.