Legal Challenges of Artificial Intelligence: An Analysis of Legal Personhood and Civil Liability for Damages Caused by Artificial Intelligence

Abstract

The rapid diffusion of artificial intelligence (AI) into medicine, transportation, finance, and public administration has exposed a structural gap between existing civil law doctrine and the practical realities of autonomous, opaque, and self-learning systems. Conventional liability regimes, namely fault-based tort, vicarious liability, and product liability, were designed around identifiable human or corporate actors and struggle whenever harm originates from an AI system's independent decision-making process. This article examines two interrelated legal problems: whether AI can or should be recognized as a subject of law, commonly termed legal personhood, and how civil liability for AI-caused damages should be allocated among developers, operators, and users. Employing a normative juridical method, the study analyzes primary legal materials, including statutory provisions, judicial doctrines, and comparative regulatory instruments such as the European Union Artificial Intelligence Act, the revised Product Liability Directive, and Indonesian civil law, alongside relevant secondary legal literature. The findings indicate that full legal personhood for AI remains conceptually premature and practically unnecessary, whereas existing liability frameworks urgently require targeted reform, particularly strict liability for high-risk applications, reversed evidentiary burdens, and mandatory insurance mechanisms. The novelty of this article lies in integrating personhood theory with liability allocation into a single risk-based accountability framework adaptable to civil law jurisdictions such as Indonesia.