Artificial Intelligence and Criminal Liability: The Incompatibility of Autonomous Decision-Making Systems with the Mens Rea Requirement

Abstract

The increasing deployment of Artificial Intelligence (AI) in autonomous decision-making systems has generated complex challenges for criminal law. Traditional criminal liability is grounded in human agency, requiring the coexistence of actus reus and mens rea. However, AI systems now operate with varying degrees of autonomy, unpredictability, and opacity, raising the fundamental question: who should bear criminal responsibility when AI causes harm? This paper examines the doctrinal compatibility of AI-related harm with existing criminal law principles, particularly within the framework of the Bharatiya Nyaya Sanhita, 2023 (BNS). Using a qualitative doctrinal and analytical legal research methodology based exclusively on statutory interpretation, judicial precedent, and contemporary scholarly literature - without reliance on empirical or experimental data - the study evaluates whether AI can be treated as a subject of criminal liability or whether responsibility must remain exclusively human-centered. The findings demonstrate that AI lacks moral agency and legal personhood, thereby preventing direct criminal attribution. Instead, liability must be distributed among developers, deployers, corporations, and regulatory authorities through a layered accountability model structured around developer responsibility, corporate governance liability, operator oversight obligations, and negligence-based attribution under existing statutory provisions. The novelty of this paper lies in integrating responsibility-gap theory with a concrete doctrinal analysis of the Bharatiya Nyaya Sanhita, 2023. Unlike prior scholarship that primarily emphasizes ethical governance or regulatory policy frameworks, this study situates AI accountability within the punitive and blame-based structure of substantive criminal law, thereby offering a legally operational model for addressing AI-related harms