Civil Liability Arising from the Design, Production, and Supply of Artificial Intelligence

Abstract

The rapid expansion of artificial intelligence and its growing application in medical, financial, commercial, industrial, and service sectors have created significant legal challenges alongside its technological and economic benefits. The complexity of algorithms, the relative autonomy of intelligent systems, their capacity for learning and behavioral modification, the involvement of multiple actors throughout the technology lifecycle, and the difficulty of identifying the direct cause of harm have exposed certain limitations of traditional civil liability rules. This study examines civil liability arising from the design, production, and supply of artificial intelligence and seeks to determine appropriate legal bases for attributing damage to the various actors involved. The research adopts a descriptive-analytical approach based on general principles of civil liability and the distinctive characteristics of AI systems. The findings indicate that designers may incur liability where harm results from unsafe architecture, unreasonable algorithmic choices, inadequate safeguards, or failure to anticipate foreseeable risks. Producers may be held responsible for defective or biased data, insufficient testing, cybersecurity weaknesses, implementation errors, inadequate quality control, or failure to provide necessary updates. At the supply stage, liability may arise from inadequate disclosure, insufficient warnings, misleading representations, inappropriate deployment, or failure to respond to risks discovered after market entry. The analysis further demonstrates that exclusive reliance on fault-based liability may be insufficient in relation to high-risk AI systems, particularly where injured parties face substantial evidentiary difficulties. Accordingly, a differentiated framework based on the level of risk, facilitated standards of proof, risk-based liability, and compulsory insurance may be necessary in specific fields. Establishing a coherent legal framework that allocates responsibility among designers, producers, suppliers, and users can enhance victim protection while simultaneously promoting the responsible development and deployment of artificial intelligence.