More than one-third of web pages published after ChatGPT's November 2022 release show signs of AI authorship or substantial AI editing, according to a study from Pew Research Center released on August 201,2.
The report used the Common Crawl web archive to collect nearly half a million English-language web pages spanning roughly five years, beginning a couple of years before ChatGPT's launch. Pew Research Center then applied Open Pangram's detection technology to assess how many pages were likely written or heavily edited by AI.
In a random sample of 10,000 web pages collected in July 2026, approximately 10% showed significant signs of AI authorship. Pew Research Center noted that a random sample of this kind inevitably includes older pages — ones published before AI writing tools existed and therefore could not have been AI-generated — which pulls the overall percentage down. When the analysis is narrowed to pages published after ChatGPT's release, the figure rises above 30%.
The study corroborates other research documenting the extent to which newer web content is now either written by or substantially edited by AI, according to Pew Research Center. TechCrunch noted that the report arrives shortly after Cloudflare reported that bot web traffic had overtaken human web traffic — a milestone Cloudflare said was reached sooner than the company had estimated.
ANALYSIS The convergence of these two data points — AI-authored content exceeding 30% of new pages and bot traffic surpassing human traffic — describes a web increasingly shaped by automated systems on both the production and consumption sides.
For AI labs and model providers, the finding carries direct implications for training-data quality: as the share of AI-generated text on the open web grows, the risk of model collapse — where models trained on synthetic outputs degrade over successive generations — becomes a more concrete operational concern.
The use of Open Pangram's classifier as the detection layer also highlights a growing market segment: tooling built specifically to distinguish human-written from AI-generated text at web scale.