Your suspicion that the internet feels increasingly artificial just got scientific backing. Stanford researchers found that by mid-2025, roughly 35% of new websites were AI-generated—a staggering leap from zero before ChatGPT’s late 2022 launch. This isn’t just a statistic about content creation; it’s evidence of the fastest technological transformation of digital space in internet history.
The Research Behind the Numbers
Researchers analyzed 33 months of archived websites to track AI content proliferation.
The study, conducted by Stanford University, Imperial College London, and the Internet Archive, examined millions of website snapshots from August 2022 to May 2025. Using Pangram v3—selected for highest accuracy among tested AI detection tools—researchers tested six hypotheses inspired by Dead Internet Theory, that dystopian concept suggesting bots and algorithms would eventually dominate online spaces. The timing wasn’t coincidental; ChatGPT’s November 2022 debut marked the inflection point where AI content creation became accessible to anyone.
Surprising Findings Challenge Doom Predictions
Only two feared outcomes materialized while misinformation concerns proved unfounded.
Despite widespread anxiety about AI flooding the web with falsehoods, researchers confirmed only two of six predicted negative impacts. AI-generated content does reduce semantic diversity—it’s less verbose and more uniform in style—while adopting an unnaturally “cheery” tone. However, human fact-checkers found no increase in false claims, and AI sites actually maintained similar citation rates to human-written ones. “The sheer speed of the AI takeover of the web is quite staggering,” noted co-author Jonáš Doležal from Stanford. “This represents a major transformation of the digital landscape.”
The Bigger Picture Gets Murkier
Supporting data shows AI content dominance extends beyond websites to publications.
By November 2024, AI-generated articles surpassed human-written ones entirely, hitting 39% of all publications according to separate research. Yet this flood of artificial content remains largely invisible in Google search results and ChatGPT responses—creating a bizarre two-tier internet where AI content proliferates but stays hidden from mainstream discovery. Stanford researchers are now developing continuous monitoring tools to track this evolution across different languages and content categories, recognizing that what started as a curiosity has become the new normal for web publishing.
