More than one-third of web pages published since ChatGPT launched show significant signs of AI authorship or editing, according to a new study by Pew Research Center.

The findings add to growing evidence that AI is reshaping not only how people use the internet, but also the content being published online. The study comes shortly after Cloudflare reported that automated bot traffic had surpassed human web traffic.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

AI Content Becomes More Common

Pew analysed nearly 500,000 English-language web pages collected through the Common Crawl archive, covering roughly five years before and after ChatGPT’s November 2022 launch. Researchers used Open Pangram’s technology to identify pages that appeared to have been written or substantially edited by AI.

A random sample of 10,000 pages collected in July 2026 showed significant signs of AI authorship on around 10% of pages. Because this sample included older content published before generative AI tools became widely available, Pew conducted a second analysis focused only on pages published after ChatGPT’s release.

That analysis found signs of AI authorship on 35% of pages.

Large Differences Across Domains

AI involvement varied considerably depending on the type of website. Pages using .com domains showed signs of AI authorship at roughly 10 times the rate seen on .edu and .gov domains, where the figure was around 1%.

Non-profit .org websites also recorded a relatively low rate, at 4.6%.

Pew cautioned that AI detection tools can incorrectly classify human-written content. However, researchers said the scale of the analysis makes the overall trend directionally meaningful.

AI Writing Patterns Are Also Changing

Alongside direct signs of AI authorship, Pew identified an increase in several writing patterns often associated with AI-generated text.

These include greater use of em dashes, Oxford commas and constructions such as “it’s not X, it’s Y.”

The findings suggest that AI is becoming an increasingly visible part of the web’s content infrastructure, even as automated systems are also becoming a larger part of the audience consuming that content.