Algorithmic Accountability Gap: Rethinking Liability Frameworks for Generative AI in Indonesian Civil Law

Abstract

Background: The rapid diffusion of generative artificial intelligence (AI) into text, image, and decision-support applications has outpaced the doctrinal assumptions of Indonesian civil liability law, which remains anchored to the fault-based tort regime of Article 1365 of the Kitab Undang-Undang Hukum Perdata (KUHPerdata). Objective: This study examines how the probabilistic, opaque, and distributed nature of generative AI strains the elements of actor, fault, causation, and foreseeability that Indonesian tort doctrine requires, and proposes a reform pathway suited to a civil law jurisdiction. Method: A qualitative doctrinal-normative legal research design was used, combining statutory and comparative analysis with a systematic thematic content analysis of forty peer-reviewed sources on algorithmic accountability, AI governance, and responsible AI practice. Results: The analysis shows that existing instruments, including the Consumer Protection Law, the Electronic Information and Transactions Law, and the Personal Data Protection Law, only partially cover generative AI harms, leaving an accountability gap between developers, deployers, and users. A tiered, presumptive liability model, expressed through a proposed Algorithmic Liability Allocation Index, is offered as a transplantable reform mechanism. Conclusion: Indonesian civil law requires a calibrated, risk-tiered liability architecture, supported by a rebuttable presumption of fault, to close the accountability gap created by generative AI without displacing the foundational logic of KUHPerdata.