인공지능(AI)의 합리적 수용을 위한 리스크 기반 규제거버넌스 ― 인공지능 기본법을 중심으로 ―

This study aims to propose a regulatory governance model for the rational social acceptance of artificial intelligence(AI). While the innovation of AI technology enhances the quality of life across various sectors, it also entails significant risks. Therefore, in order to build public trust in AI, it is necessary to devise balanced regulatory measures that maximize benefits while minimizing adverse effects. Traditional regulatory approaches, however, may hinder technological innovation and fail to achieve the intended regulatory objectives. To address these limitations, this study explores a governance framework centered on the “Framework Act on Artificial Intelligence,” with the goal of contributing to the establishment of a trustworthy AI ecosystem. First, this paper examines the normative foundations of regulatory governance as a diversified alternative to the limitations of existing regulatory systems. Governance is defined as the organic development of regulation through interactions among various actors, thereby embodying regulatory pluralism. The study seeks to integrate governance into the legal discourse by theoretically analyzing it as a method for overcoming the limitations of conventional regulation. Second, by comparing different regulatory approaches adopted by major AI-developing countries, the paper derives insights applicable to Korea. Countries employ hybrid regulatory models tailored to their domestic context—ranging from combinations of self-regulation and principle-based approaches to more stringent ex-ante regulations using risk-based and rule-based methods. The paper categorizes regulatory approaches such as risk-based regulation, principle- or rule-based models, and horizontal or sector-specific regulation, offering a comparative analysis of Korea’s regulatory model. Third, the study proposes a multi-layered regulatory governance model that integrates public law-based legal regulation with private sector-led self-regulation. By identifying the structural limitations of the current Framework Act on Artificial Intelligence, especially regarding its risk-based approach, it suggests improvements through a combination of rule-based and self-regulatory mechanisms to better manage and prevent risks. Furthermore, it discusses the necessary institutional arrangements and supervisory authorities, while also presenting strategies for harmonizing with global regulatory frameworks. It is expected that the proposed regulatory governance model will serve as a normative response to enhance the rational acceptance of AI and promote its broader societal integration.