CETaS Paper – From Cybercrime to Vibercrime? (cetas.turing.ac.uk)

A recent study from researchers at Cambridge, Strathclyde, and Edinburgh Universities explores the implications of generative AI on the cybercrime landscape, particularly focusing on how low-level cybercriminals engage with these technologies. Analyzing over 100 million messages from underground forums, the research reveals a nuanced picture: while there is excitement around "dark AI" tools tailored for cybercrime, their practical impact remains limited. Many criminals lack the technical skills to effectively utilize advanced AI models like ChatGPT and Claude, with discussions highlighting that innovative uses for these tools in crime are still in their infancy. The implications for the AI/ML community are significant, shedding light on how generative AI's automation capabilities could enhance certain types of cybercrime, particularly in scaling operations like romance fraud. However, the study also notes saturation effects, where the market for targets becomes limited, and the increased use of AI may lead skilled IT workers to consider illegal activities after job losses due to AI adoption in legitimate sectors. The findings indicate that while generative AI is being integrated into cybercriminal practices, the broader cultural tensions and ethical concerns within these communities reflect a struggle between the adoption of new technology and the preservation of traditional hacker skills and values.