AI 거버넌스의 새로운 축(軸) ― 영향평가제도에 관한 연구 ―

As digital transformation accelerates, Data, Platform, and Artificial Intelligence (hereinafter “DPA”) are forming an interconnected ecosystem and becoming the foundation of modern society. In particular, AI systems, as a pivotal component of the DPA ecosystem, create innovative value while simultaneously generating new risks such as data bias, algorithmic discrimination, and digital exclusion. In this context, impact assessment systems are gaining attention as a new axis of AI governance and as a key regulatory tool that can effectively control negative externalities without hindering AI systems' innovative potential. This article theoretically examines the relationship between risk assessment and impact assessment, and conducts a comparative legal analysis of AI impact assessment legislation in the EU, US, Canada, and Korea. Through this analysis, it confirms that impact assessment systems bring about a paradigm shift in AI regulation and identifies two key issues: the absence of a universal assessment model and the relationship between risk and impact. Based on this analysis, three directions are proposed for designing Korea's AI impact assessment system: first, establishing an assessment framework that integrates risk and impact considerations; second, adopting a phased and differentiated approach; and third, building an effective governance system. This article anticipates that its analysis and recommendations can contribute to establishing a new regulatory governance system commensurate with the digital convergence era.