AI's Growing Footprint: How the 'Dead Internet' Theory Gains Credibility from Stanford Research
The Dead Internet Theory: From Conspiracy to Reality
For years, fringe theorists have whispered about a 'dead internet'—a digital realm where human voices are drowned out by automated bots and algorithmically generated content. Once dismissed as paranoia, this concept is now gaining empirical support. A new study from Stanford University, Imperial College London, and the Internet Archive reveals that over a third of all new websites are AI-generated or AI-assisted, marking a rapid shift in the online landscape since 2022.

What the Research Reveals
To quantify AI's infiltration, researchers turned to the Wayback Machine, comparing web pages published between 2022 and 2025. Using multiple AI-detection methods, they analyzed millions of pages. The findings are stark: by May 2025, 35.3% of all new websites were either entirely or partially produced by artificial intelligence. Of that, 17.6% were completely AI-generated. This dramatic uptick, concentrated in just three years, suggests that the dead internet theory is no longer speculative—it's measurable.
Corroborating Evidence from the Web
The Stanford study aligns with other industry data. Cloudflare reported that nearly a third of all internet traffic in the past year came from bots. Similarly, Imperva found that automated traffic surpassed human traffic for the first time in 2024. These numbers paint a consistent picture: machines are not just visitors—they're becoming the primary inhabitants of the web.
Beyond the Hype: AI's Surprising Impact
While the prevalence of AI-generated content might sound alarming, the study challenges some common fears. The researchers tested six hypotheses about negative effects. Two were confirmed: AI contributes to semantic contraction (reducing diverse viewpoints) and a positivity shift (making online writing more sanitized and artificially cheerful). However, four feared outcomes were not observed.
What AI Hasn't Done (Yet)
Contrary to popular belief, the study found no evidence that AI-generated text leads to rambling, substance-free content. Nor did it find a homogenization of writing styles, a lack of cited sources, or—most surprisingly—a widespread spread of misinformation through AI-produced articles. This suggests that while AI is reshaping the web, it may not be as toxic as many assume.
A Transformative Shift in Digital Landscape
Jonáš Doležal, a researcher on the study, commented, "I find the sheer speed of the AI takeover of the web quite staggering. After decades of humans shaping it, a significant portion of the internet has become defined by AI in just three years." The researchers are now working on a continuous monitoring tool to track these trends in real time. As the dead internet theory edges closer to reality, this research offers both a warning and a nuanced perspective: the internet is changing, but not always in the ways critics expect.
Related Articles
- Unlocking the Cambrian Explosion: What New Fossil Discoveries Reveal About Early Animal Evolution
- How to Deploy AI Agents with Secure Desktop Access Using Amazon WorkSpaces
- Automated Failure Attribution in LLM Multi-Agent Systems: A Practical Guide to Identifying Responsibility for Task Failures
- Andrew Forrest Declares War on Fossil Fuels: Fortescue to Lead Australia to 'Real Zero'
- The Stealthy Saboteur: Inside the Fast16 Malware Campaign
- How the First Letter of RNA Shapes Antiviral Alarms: A vs G
- How Alibaba's Metis Agent Slashes Unnecessary AI Tool Use by 96% While Boosting Accuracy
- Atlantic Storms Disrupt African Dust Flow to Amazon, Study Warns