블랙박스 AI 의료에 대한 유럽연합과 미국 불법행위법의 대응

EU and US tort law responses to black box AI medical injuries

This paper aims to identify the characteristics of “black box” AI medicine and, based on several hypothetical scenarios, to clarify the attribution of liability between AI manufacturers, healthcare providers, and consumers (patients). The opacity of AI medicine, which makes the allocation of liability for medical injuries to patients unclear, is a pressing issue. The revised EU Product Liability Directive, which came into effect on December 8, 2024, has made it easier to seek redress for damages caused by AI products. The revised Directive includes software and AI within the definition of “products” and significantly eases the burden of proof for victims through disclosure orders and the presumption of negligence and causation. These legislative measures are considered to have significantly eased the difficulties of seeking redress for damages caused by “black box” AI medicine. There is a perception that US tort law and case law related to product liability have limitations in pursuing legal liability for injuries caused by AI medicine. This perception calls for assimilation with the EU Product Liability Directive. South Korea's product liability law is overly simplified and has limitations in addressing liability issues arising from AI-based medical devices. The EU Product Liability Directive and legislative discussions in the US could provide important insights for reforming product liability law.