CIVIL LAW AND CIVIL PROCEDURE

Civil Liability and Risk Sharing Mechanisms for Harm Caused by Artificial Intelligence

Abstract

This article provides a comprehensive analysis of civil liability for harm caused by artificial intelligence systems. The relevance of the study is determined by the rapid integration of AI technologies into all spheres of public life in the absence of specific provisions in Russian legislation governing compensation for damage caused by such systems. It is argued that the technical characteristics of AI – namely autonomy, the capacity for self-learning with dynamically evolving architecture, the probabilistic nature of decision-making, and structural complexity – significantly complicate the application of traditional doctrines of civil liability, particularly with respect to establishing causation and identifying the proper liable party.

On the basis of an analysis of doctrinal approaches and regulatory legal acts, the study identifies legally significant features of AI that influence the choice of an appropriate liability regime. A differentiated approach to models of civil liability is proposed, depending on the level of risk and the sphere of application of AI systems: strict (no-fault) liability of developers and operators is suggested for high-risk systems, whereas fault-based liability of the tortfeasor is deemed appropriate for low-risk systems. In order to mitigate the adverse consequences of harm and ensure effective compensation for damages, it is proposed to introduce legislative provisions enabling the establishment of self-regulatory organizations of AI developers with mandatory safety standards and compensation funds, to implement compulsory civil liability insurance, and to create a specialized state compensation fund financed through contributions from producers of highly autonomous systems.