started · updated
Pew Research: AI-generated content exceeds 35% of new web pages
A study by the Pew Research Center reveals a significant increase in AI-generated or AI-edited content across the internet. After analyzing nearly 500,000 English-language webpages using the Open Pangram detection tool, researchers found that approximately 10% of all examined pages in July 2026 showed signs of AI involvement.
The impact is even more pronounced when looking at recent content. For webpages published after the release of ChatGPT in late 2022, the share of pages showing signs of AI authorship or editing rises to over 35%. The distribution of this content is uneven across different web domains; commercial .com sites show signs of AI usage at roughly ten times the rate of .edu or .gov domains, which both hover around 1%.
The study also highlights shifts in online linguistic patterns. Since 2023, the frequency of em dashes has doubled, and the use of the Oxford comma has increased by 63%. Additionally, the usage of specific words frequently favored by AI models, such as 'delve' and 'interplay', has more than doubled. The research indicates that while these markers cannot definitively prove a single document is AI-generated, they demonstrate a clear trend in how language is being used across the web.
Entities
ChatGPT · Common Crawl · Open Pangram · OpenAI · Pew Research Center
Claims
What the coverage asserts, and how many sources carry each claim.
- [DISPUTED] More than one-third of webpages published after the release of ChatGPT show signs of AI authorship or editing. itreseller.pl · universoabierto.org · slashdot.org · assodigitale.it · the-decoder.com · +1 more
- [● 6 SOURCES] AI-generated content is significantly more prevalent in .com domains than in .edu or .gov domains. itreseller.pl · universoabierto.org · slashdot.org · assodigitale.it · the-decoder.com · +1 more
- [● 4 SOURCES] The usage of specific words frequently associated with AI, such as 'delve' and 'interplay', has more than doubled. itreseller.pl · slashdot.org · the-decoder.com · www.searchenginejournal.com
- [● 4 SOURCES] The use of the Oxford comma on the web has increased by 63% since 2023. itreseller.pl · slashdot.org · the-decoder.com · www.searchenginejournal.com
- [● 4 SOURCES] The frequency of em dashes in online text has approximately doubled since 2023. itreseller.pl · slashdot.org · the-decoder.com · www.searchenginejournal.com
- [● 6 SOURCES] Approximately 10% of all analyzed English-language webpages showed signs of AI authorship or editing as of July 2026. itreseller.pl · universoabierto.org · slashdot.org · assodigitale.it · the-decoder.com · +1 more