The Responsibility Gap in Algorithmic Governance:A Legal Analysis of Criminal Liability and Due Process in Automated Administrative Decisions.

Abstract

The accelerating integration of automated decision-making (ADM) systems into public administration has generated a profound accountability deficit that existing legal frameworks are inadequately equipped to address. This paper examines the 'responsibility gap'—the structural absence of clearly assignable liability when algorithmically driven decisions produce legally harmful outcomes in administrative and judicial contexts. Drawing on a normative juridical methodology and comparative regulatory analysis, the study interrogates the doctrinal coherence of established criminal law concepts, particularly Actus Reus and Mens Rea, when applied to harms caused by autonomous computational agents. The analysis critically evaluates recent legislative instruments, including the EU Artificial Intelligence Act (2024), the General Data Protection Regulation, and Saudi Arabia's Personal Data Protection Law (PDPL), to assess the feasibility of attributing criminal and administrative liability to non-human entities. The paper advances three original contributions: a Hybrid Liability Model that distributes responsibility across the algorithmic supply chain; a Legal Risk Index (LRI) as a standardised diagnostic instrument for regulatory compliance; and a constitutional reframing of Explainable AI (XAI) as an enforceable component of procedural due process. The findings demonstrate that effective governance of algorithmic systems requires not merely technical solutions but a coherent jurisprudential architecture capable of preserving the rule of law in an increasingly automated state