Study: One-Third of New Websites Are AI-Generated Since ChatGPT's Launch
Key Takeaways
- ▸35% of websites created by mid-2025 are AI-generated or AI-assisted, a transformation from 0% before ChatGPT's late 2022 launch
- ▸AI-generated content makes the web more positive and optimistic but less semantically diverse
- ▸Contrary to widespread fears, the study found no significant increase in verifiable misinformation or reduction in source citations
Summary
Researchers from Stanford, Imperial College London, and the Internet Archive have published findings showing that approximately 35% of newly created websites by mid-2025 were AI-generated or AI-assisted—up from essentially zero before ChatGPT's launch in late 2022. The study, titled "The Impact of AI-Generated Text on the Internet," analyzed 33 months of web snapshots to quantify how rapidly generative AI has reshaped the internet's content landscape.
The research tested six popular theories about AI-generated text's negative effects: increasing misinformation, reducing viewpoint diversity, eliminating source citations, lowering semantic quality, reducing writing substance, and creating stylistic uniformity. Using AI-detection software (Pangram v3) and human fact-checkers, researchers found surprisingly nuanced results that challenge prevailing fears. While AI content did make the internet "more cheery and less semantically diverse," the study found no evidence of widespread verifiable falsehoods or reduced source linking—contradicting the popular "Dead Internet Theory."
Researcher Jonáš Doležal from Stanford emphasized the speed of transformation: "After decades of humans shaping it, a significant portion of the internet has become defined by AI in just three years." The findings suggest AI's impact is more complex than commonly feared, with both genuine concerns about diversity and unexpected resilience against misinformation.
- Researchers identified a potential blind spot: AI may be increasing unverifiable claims rather than outright falsehoods
Editorial Opinion
This research offers essential nuance to the often-alarmist discourse around AI's internet impact. While 35% AI-generated websites is undoubtedly striking, the absence of a misinformation explosion is reassuring and worth celebrating. However, the observed shift toward semantic uniformity and the researchers' caveat about unverifiable claims warrant serious ongoing attention—the internet's diversity and epistemic health extend beyond simple fact-checking.
