The Structural Dilemma of AI-Driven Cybercrime

AI-driven cybercrime has led to a dilution of criminal responsibility along the chain of attribution. This dilution manifests at three levels: the spatiotemporal separation between conduct and consequence at the causal level; the opacity of causal chains at the subjective level; and the pluralization of responsible subjects at the attributable level. The traditional framework of criminal liability, grounded in the subjective element, objective element, and subject qualification, faces significant evidentiary challenges in the context of AI-driven cybercrime. Mainstream criminological theories have not entirely lost their explanatory power; however, certain underlying assumptions require targeted adjustments. These include the deterrence target presupposed by rational choice theory, and the guardianship capacity embedded in routine activity theory. This article proposes an analytical framework centered on the “dilution of responsibility” and, on this basis, advances a multi-tiered governance scheme. At the level of criminal law, it suggests introducing an algorithmic product liability regime modeled on product liability doctrine. At the administrative law level, it advocates for the establishment of mandatory security assessment mechanisms. In civil law, it proposes no-fault liability remedies for victims. At the level of social governance, it emphasizes AI literacy as a foundational means of reducing victim vulnerability. The theoretical contribution of this study lies in conceptualizing the impact of artificial intelligence on cybercrime as a structural dilution of responsibility rather than a fundamental transformation of the nature of crime, thereby providing constructive insights for both theoretical analysis and institutional design.