Introduction

The concept of a "dead internet"—where human-driven content has been supplanted by automated bots—has moved from fringe conspiracy theory to a measurable reality. A new study led by researchers at Stanford University, Imperial College London, and the Internet Archive provides the first comprehensive calculation of just how pervasive AI-generated text has become online. The findings reveal that more than a third of all new websites are now created with artificial intelligence, marking a dramatic shift in the digital landscape.

How the Study Measured AI Domination

To assess the extent of AI-generated content, the team analyzed web pages archived by the Wayback Machine, comparing pages published from 2022 to 2025. They employed multiple AI-detection methods to distinguish human-written text from machine-generated or machine-assisted content.

Key Findings

As of May 2025, the data shows that 35.3% of all newly published websites were either entirely generated by AI or relied on AI assistance. Notably, 17.6% of these sites were fully AI-authored—with no direct human input in the text.

These numbers align with other industry reports. Cloudflare has noted that nearly a third of all internet traffic over the past year originated from bots, while Imperva reported that in 2024, automated traffic exceeded human traffic for the first time in history.

"I find the sheer speed of the AI takeover of the web quite staggering," said Jonáš Doležal, one of the study's researchers, in an interview with 404 Media. "After decades of humans shaping it, a significant portion of the internet has become defined by AI in just three years. We're witnessing, in my opinion, a major transformation of the digital landscape in a fraction of the time it took to build in the first place."

Not All Effects Are Negative: Surprising Results

While the rapid growth of AI-generated content raises concerns, the study also investigated whether common criticisms of such content hold true. The researchers tested six hypotheses about the negative impacts of AI text—and only two were confirmed.

Confirmed: Semantic Contraction and Positivity Shift

The study found evidence of semantic contraction, meaning a reduction in the diversity of viewpoints and vocabulary online. Additionally, a positivity shift was observed, where online writing has become more sanitized and artificially cheerful.

What AI-Generated Content Hasn't Done (Yet)

Surprisingly, the research did not find evidence that AI text leads to:

• Rambling, low-substance writing
• A single, generic style across all AI content
• A lack of cited sources
• Increased spread of misinformation

This suggests that the worst fears about AI degrading the internet's quality may be premature—or that current AI-generated content is more sophisticated than critics assume.

Toward Continuous Monitoring

The research team plans to turn their methodology into a continuous tracking tool, allowing internet users and researchers to monitor the evolving presence of AI-generated text in real time. As the technology advances, staying informed will be crucial for understanding how the "dead internet" theory continues to develop.

Conclusion

While the internet is far from entirely dead, the data shows a clear and rapid takeover by AI-generated content. The study offers both a wake-up call and a nuanced perspective: AI is reshaping the web at an astonishing pace, but not every predicted negative effect has materialized. As we move forward, balancing the benefits of AI assistance with the need for authentic human expression will remain a central challenge.