AI-written webpages are becoming a significant part of the modern internet, with more than one-third of English-language webpages published since ChatGPT launched showing signs of being written or substantially edited by artificial intelligence. A new analysis from Pew Research Center suggests that generative AI is not simply changing how people create content but is also beginning to influence the language and writing style found across the web.
AI Content Has Grown Rapidly
Pew analyzed around 490,000 English-language webpages collected through Common Crawl between January 2021 and July 2026. Researchers used the Open Pangram AI-detection model to identify patterns associated with artificial intelligence.
When the entire July 2026 sample was examined, including webpages published before the emergence of modern generative AI, about 10% showed significant signs of AI authorship.
The picture changed considerably when researchers focused only on pages published after ChatGPT’s launch in November 2022. In that group, more than one-third showed significant AI signals.
The findings indicate how quickly AI-written webpages have become part of the online publishing ecosystem.
Commercial Websites Lead AI Adoption
AI involvement is not evenly distributed across different types of websites.
Pew found that approximately one in 10 webpages on .com domains showed signs of AI authorship in its 2026 samples. That rate was about twice as high as the 4.6% recorded on .org websites.
Government and educational websites showed substantially lower levels. Only around 1% of webpages on both .edu and .gov domains displayed significant signs of AI authorship.
The differences could reflect how commercial websites have more incentives to produce large volumes of content quickly. Businesses, publishers and online service providers can use AI tools to generate drafts, rewrite existing material, create product descriptions and produce other forms of digital content.
AI Is Changing Online Writing Styles
The influence of AI may extend beyond the question of who writes an article.
Pew identified several writing patterns that have become more common online since 2023. One notable example is the use of em dashes, which has roughly doubled.
Oxford commas have also increased by 63%, while certain words frequently associated with AI-generated writing have become much more common.
Terms such as “delve,” “interplay” and “testament” have more than doubled in frequency, according to the research.
Another writing pattern, known as negative parallelism, has also increased significantly. This includes constructions such as “it’s not just X, it’s Y.” Pew found that this pattern has nearly tripled, although it remains relatively uncommon compared with ordinary writing structures.
These trends suggest that AI-written webpages may be contributing to a gradual change in the overall tone and vocabulary of online content.
Detection Is Not Perfect
The findings should not be interpreted as proof that every webpage identified by the analysis was produced by AI.
Pew emphasized that individual writing habits cannot reliably establish whether a specific article was generated by artificial intelligence. Human writers can naturally use em dashes, Oxford commas and words such as “delve.” Similarly, AI-detection systems can incorrectly classify both human-written and machine-generated material.
This is an important distinction because AI detection remains an imperfect process.
Instead of determining whether a particular author used AI, large-scale analysis can reveal broader patterns. When hundreds of thousands of webpages are examined together, recurring changes become easier to identify.
What This Means for Online Publishers
The rise of AI-written webpages presents both opportunities and challenges for publishers.
AI can help content teams work more efficiently, particularly when they need to create large amounts of material. Editors can use these systems for brainstorming, summarizing information, improving grammar or producing initial drafts.
However, simply increasing publishing volume does not necessarily improve the quality of the internet.
As AI-generated material becomes more common, publishers may need to place greater emphasis on original reporting, expert knowledge, fact-checking and editorial oversight. Human judgment remains particularly important when content involves complex subjects, breaking news or claims that require verification.
Websites that publish large quantities of generic AI-generated material may also struggle to distinguish themselves from competitors producing similar content.
A New Era for Digital Content
The growing presence of AI-written webpages reflects a broader transformation in how information is produced and distributed.
Since ChatGPT’s arrival, generative AI has moved rapidly from an experimental technology to a mainstream tool used by writers, businesses, developers and publishers. The Pew findings provide an indication of how deeply that transition is already affecting the web.
At the same time, the research highlights the difficulty of separating human and machine writing. As AI systems become more sophisticated and people increasingly edit machine-generated drafts, the boundary between AI-assisted and human-created content is becoming less clear.
The future of online publishing may not be defined simply by whether content is written by a person or an AI system. Instead, the more important question may be whether the information is accurate, useful, original and responsibly edited.
The growth of AI-written webpages makes that standard increasingly important as the internet enters a new phase of AI-assisted publishing.
