Published Paper Details:

The rapid diffusion of generative artificial intelligence (GenAI) has fundamentally altered the architecture of criminal harm by enabling scalable, low-cost, and highly realistic forms of digital misconduct. While existing legal frameworks nominally extend to such harms, they remain conceptually fragmented and doctrinally strained when applied to AI-mediated conduct. This article addresses this gap by developing a structured taxonomy of GenAI-enabled criminality, distinguishing between misinformation, fraud and identity theft, hate speech, and sexual offences, while recognizing deepfakes as a cross-cutting modality. Adopting a comparative doctrinal approach, the study analyses regulatory and criminal law responses in the United States, the European Union, and India. It argues that contemporary legal systems rely excessively on analogical extension of pre-existing offences. The article advances a taxonomy-driven framework for legal classification and proposes a recalibration of liability principles to address intent, agency, and harm in AI-mediated environments. It concludes that without doctrinal refinement and targeted regulatory intervention, criminal law risks both under-inclusivity and overreach in responding to synthetic harms.