의료 분야에서 인공지능 활용의 법적 쟁점 – 법적 인격을 중심으로 –

In the medical field, artificial intelligence (AI) is expected to enhance diagnostic accuracy and address regional and social disparities. However, AI also raises significant issues, such as data bias, algorithmic opacity, and unclear responsibility. This paper first identifies the general problems associated with AI and provides an overview of how the safety and efficacy of AI medical devices are being verified under current medical law and administrative practices. While regulatory frameworks for the performance of AI medical devices are somewhat established, the civil liability issues related to their use are not specifically addressed within medical laws and are instead governed by general civil law principles. At the current level of AI development, issues related to civil liability for AI medical devices have not yet become prominent. However, as AI evolves and becomes more integrated into medical practice, the reliance on strong AI could complicate the assignment of liability for medical malpractice. This paper examines how civil liability for medical malpractice involving advanced AI medical devices might be regulated under current laws and explores the potential limitations of such regulations. Approaches to this issue generally fall into two categories: the first approach seeks to reinforce the existing principles of civil liability by enhancing the responsibilities of manufacturers (providers) and users, while the second approach proposes granting legal personhood to AI, making AI entities themselves responsible for their actions. As AI medical devices become more advanced and widely used, the traditional approach of treating AI as a tool and holding manufacturers or users liable may reveal its limitations. Consequently, this paper emphasizes the argument for granting legal personhood to AI medical devices. The latter approach, however, requires caution to prevent the deflection of responsibility through the granting of legal personhood and must be designed to preserve the human-centric foundation of current legal systems. Therefore, this paper presents fundamental principles that should be adhered to when considering the legal personhood of AI.