Key facts
- Over one-third of web pages published after ChatGPT's release show signs of AI authorship.
- A Pew Research study found 35% of post-ChatGPT pages had significant AI authorship signs.
- .com domains showed AI authorship at 10 times the rate of .edu or .gov domains.
- AI detection tools can misclassify pages, but data is directionally correct.
- Increased use of em dashes and Oxford commas were noted as AI authorship tells.
A study by Pew Research released on Thursday indicates that over one-third of web pages published since the launch of ChatGPT exhibit signs of AI authorship or substantial editing. This finding aligns with other research suggesting a significant increase in AI-generated content online.
The report utilized data from the Common Crawl web archive, analyzing nearly half a million English-language web pages. Using Open Pangram's technology, Pew found that in a sample of 10,000 pages collected in July 2026, approximately 10% showed significant signs of AI authorship. However, to better assess the impact of recent AI tools, Pew filtered out older pages published before AI writing tools were widely available.
Focusing exclusively on pages published after ChatGPT's November 2022 release, the study revealed that 35% displayed significant indicators of AI authorship. The analysis also noted domain-specific trends, with .com domains showing AI authorship at a rate approximately 10 times higher than .edu or .gov domains, which had a 1% rate. .org domains had a 4.6% rate of AI authorship.
Pew acknowledged that AI detection tools are not perfect and can sometimes misclassify human-written content. However, the firm stated that the data is likely directionally accurate at scale. The study also pointed to an increase in stylistic markers often associated with AI, such as the use of em dashes, Oxford commas, and certain phrasing patterns.