AI-Generated Web Pages Account for 35% of New Content

A new study has found that AI-generated web pages are becoming common across the internet. Pew Research Center found clear signs of AI involvement in about 35 percent of new web pages. 

The findings come as researchers examine how automated systems are changing online content. The study analysed a web archive containing nearly 500,000 English-language web pages. These pages were collected over the past five years for the research. 

Researchers examined the text using an OpenAI tool called OpenAI’s classifier. The tool estimates the likelihood that artificial intelligence contributed to written content. Across the full sample, about 10 percent of pages showed clear signs of AI-generated writing. 

However, the figure increased to 35 percent among pages published after ChatGPT was launched. The difference indicates a significant rise in AI involvement in newly published online material. The findings also relate to wider discussions about the so-called “dead internet theory”. 

The theory suggests automated systems and bots could increasingly shape online content and consumption. AI-generated web pages now form a growing part of this changing digital environment. Researchers say the expansion of AI tools could make online content harder to classify. 

AI-generated web pages may also appear alongside human-written material without clear labels. This creates a concern about how readers identify the source of information online. 

The study does not claim that every page showing AI signals was fully written by AI. Instead, the analysis identifies pages with signs of substantial AI involvement. The research provides data on how quickly AI is entering everyday web publishing. 

As AI tools become more widely available, their role in content creation is likely to receive greater scrutiny. The growing presence of AI-generated web pages could make distinguishing human and machine-created content more difficult. 

For internet users, this may increase worry about understanding how online information is produced. The study highlights a rapidly changing web where human and automated content increasingly exist together.