Emerging technologies, including artificial intelligence (AI), are rapidly outpacing traditional legal frameworks, exposing regulatory gaps and weakening the effectiveness of conventional governance mechanisms. This study examines liability gaps stemming from algorithmic opacity, the distributed architecture of AI systems, and the systemic and diffuse nature of AI-related harms. These characteristics highlight the inadequacy of both fault-based and strict liability regimes in addressing AI-specific risks. The analysis is situated within the context of regulatory sandboxes, adaptive governance instruments designed to balance innovation and risk management by permitting firms to test new technologies under regulatory supervision with temporary legal exemptions. While such models effectively foster innovation, they leave unresolved questions of liability when harms arise from compliant experimentation. By placing primary civil and criminal responsibility on participating firms, sandbox frameworks fail to account for the distinctive nature of AI-related harms, resulting in compensation mechanisms that are often insufficient, delayed, and burdensome for affected parties. To address this gap, the study advances the proposal of a state-backed no-fault compensation scheme, modeled on the Vaccine Injury Compensation Programs (VICPs).
Hien Tran (2025) studied this question.