Shifting Liability Principles for Generative Artificial Intelligence from a Law and Economics Perspective: From Negligence Liability to Strict Liability

Abstract

While generative artificial intelligence's technological breakthroughs unleash tremendous value, its unique tort risk structure fundamentally challenges traditional negligence liability principles. This thesis systematically analyzes, from a law and economics perspective, the inherent defects and institutional failure origins of applying negligence liability principles to Generative AI infringements, and demonstrates the rationality and advantages of shifting toward strict liability principles. Through typological analysis of Generative AI infringement scenarios, four core characteristics are distilled: technological black box and causal concealment, systemic risk diffusion, web-like dissolution of responsible entities, and inevitable damages with high remediation costs. The thesis indicates that negligence liability principles face systemic failures including judicial difficulties in determining reasonable care standards, excessive expansion of behavioral standards, and inherent defects in alternative attribution schemes. In contrast, strict liability principles demonstrate significant advantages by simplifying attribution requirements, optimizing care level incentives, and internalizing activity level costs. Therefore, this thesis argues that under existing legal frameworks and technological conditions, strict liability principles represent a more feasible and necessary institutional choice.