CIVIL LIABILITY FOR ARTIFICIAL INTELLIGENCE-BASED DECISION MAKING: LEGAL CHALLENGES AND FUTURE DIRECTIONS

Abstract

The rapid diffusion of artificial intelligence (AI) into decision-making processes across banking, healthcare, transportation, and public administration has exposed significant gaps in existing civil liability regimes, most of which were designed for human tortfeasors or physical products rather than autonomous, self-learning systems.^[1] This article examines the legal challenges that AI-based decision-making poses to conventional doctrines of fault, causation, and foreseeability, and evaluates the extent to which existing civil law frameworks, both in comparative jurisdictions and in Indonesia, remain adequate to allocate liability when an algorithmic decision causes harm. Using a normative juridical method combining statutory, comparative, and conceptual approaches, the study analyzes recent regulatory developments, including the European Union’s now-withdrawn AI Liability Directive, the revised Product Liability Directive, and Indonesia’s Civil Code, Consumer Protection Law, and Personal Data Protection Law. The findings indicate that the opacity of many AI systems, the diffusion of responsibility among developers, deployers, and users, and the unresolved question of AI legal personhood collectively produce what several scholars describe as a liability gap that existing fault-based regimes cannot adequately close. The article concludes by proposing a layered civil liability framework that combines risk-based strict liability, a rebuttable presumption of causation, mandatory insurance mechanisms, and algorithmic transparency obligations as future directions for both international and Indonesian law reform, and it argues that legislative intervention, rather than incremental judicial adaptation alone, is necessary to restore legal certainty for victims of AI-related harm.