Legal and Comparative Analysis of Civil Liability of Artificial Intelligence in Automated Decision-Making

Abstract

With the rapid advancement of human-centric artificial intelligence, this technology has become a significant factor influencing various fields, including medicine, industry, transportation, commerce, law, and banking. Machine-based artificial intelligence, through automated decision-making, has taken over tasks that were previously performed by humans. However, a fundamental challenge in this area is determining legal liability in cases where AI systems make errors. The significance of this issue lies in the fact that, in many countries, traditional civil liability laws are primarily based on human will and actions, and a comprehensive legal framework for artificial intelligence and automated decision-making has yet to be developed. This paper examines the theoretical foundations of civil liability in automated decision-making and the emerging challenges in this domain. This study employs an analytical-descriptive and comparative legal research method. Different legal systems have adopted varying approaches to determining AI civil liability. In the United States, liability is primarily assessed under product liability and vicarious liability doctrines, whereas the European Union is moving towards a strict liability model and AI civil liability insurance. In Iran, civil liability remains based on human and corporate legal personality, and AI lacks independent legal personality. Some legal systems, such as Germany, have proposed that AI-based decision-making should be subject to a corporate liability model. This paper analyzes various models for determining AI civil liability, including developer and manufacturer liability, operator liability, strict liability, civil liability insurance, and the possibility of granting AI limited legal personality. The strict liability model, which is gaining traction in the European Union, holds organizations accountable for AI-related damages, regardless of fault. In Iran, the absence of specific regulations in this area may lead to significant legal ambiguities and enforcement challenges in AI-related lawsuits. Finally, this paper emphasizes the necessity of drafting new legal regulations aligned with international standards to effectively address the legal challenges and civil liability issues arising from AI-driven automated decision-making.