A recent study by Stanford University, Imperial College London, and the Internet Archive forecasts that by mid-2025, 35% of new websites will be AI-generated or AI-assisted. This marks a significant increase from nearly zero before the launch of ChatGPT in November 2022. The research, titled "The Impact of AI-Generated Text on the Internet," utilized data from the Internet Archive and classified web pages using the AI text detector Pangram v3. The study highlights two key findings: AI-generated websites exhibit 33% higher semantic similarity compared to human-written content, indicating a decrease in semantic diversity. Additionally, AI-generated content scores 107% higher in positive sentiment, attributed to language models' tendency to produce sanitized and upbeat text. Despite concerns, the study found no significant evidence that AI-generated content reduces factual accuracy on the internet. Researchers are now working with the Internet Archive to develop a real-time monitoring tool to track AI's presence online.