AI Now Powers Over a Third of New Websites, Landmark Study Reveals

Breaking: AI-Generated Content Surges to 35% of New Websites

More than one-third of all new websites are now created with artificial intelligence, according to a groundbreaking study from Stanford University, Imperial College London, and the Internet Archive. The research, which analyzed web pages published between 2022 and 2025 using the Wayback Machine, found that 35.3% of new sites are AI-generated or AI-assisted—with 17.6% entirely produced by AI.

These findings mark a dramatic shift in the digital landscape, fueling concerns that the long-discussed 'dead internet' theory is becoming a reality. The theory posits that human-created content is being systematically replaced by bots and AI-generated material, often to manipulate public perception.

Study Methodology

The team employed multiple AI-detection tools to compare web pages over time, tracking the explosion of synthetic text since ChatGPT‘s debut in 2022. Their results corroborate earlier reports: Cloudflare found that nearly a third of all internet traffic now comes from bots, and Imperva confirmed that automated traffic surpassed human traffic for the first time in 2024.

Expert Reaction

“I find the sheer speed of the AI takeover of the web quite staggering,” said Jonáš Doležal, one of the study’s researchers. “After decades of humans shaping it, a significant portion of the internet has become defined by AI in just three years.”

Background: The 'Dead Internet' Theory

The so-called 'dead internet' theory has long been a topic among conspiracy theorists, who argue that online spaces—once filled with human voices—are now dominated by bots posing as people. While extreme versions allege deliberate manipulation by governments and corporations, the new data suggests a more organic but equally alarming trend: the rapid automation of content creation.

What This Means

The study tested six common criticisms of AI-generated text, but only two were confirmed. Semantic contraction—the reduction of diverse viewpoints—is evident, as is a positivity shift that makes online writing more sanitized and artificially cheerful.

However, researchers found no evidence of increased rambling, a single generic style, lack of cited sources, or—surprisingly—a rise in misinformation from AI-generated content. “Defying AI stereotypes, the worst fears haven’t materialized yet,” the authors noted.

Implications for Users

• AI content is now ubiquitous, making it harder to distinguish human from machine writing.
• While misinformation hasn‘t spiked, the narrowing of expression could stifle authentic discourse.
• The research team is developing a continuous monitoring tool to track these trends in real time.

For now, the internet is not dead—but it is being rapidly reshaped by AI. Back to top