금융부문 AI 규제 체계에 관한 연구: EU의 입법례를 중심으로

This study provides an in-depth analysis of the EU's regulatory framework for AI in the financial sector, examining the regulatory challenges that arise at the intersection of AI's universality and the financial sector's specificities. The EU's multi-layered regulatory approach, based on the AI Act and complemented by the Digital Operational Resilience Act (DORA) and the Financial Data Access framework (FIDA), represents a pioneering attempt to harmoniously balance the innovativeness of AI technology with the stability of the financial industry. The research reveals that the EU's financial AI regulatory framework encompasses the following distinctive elements: First, it establishes an integrated regulatory framework that organically links the horizontal regulation of the AI Act with vertical regulations reflecting the specificities of the financial sector, such as DORA and FIDA. This approach is noteworthy for enhancing the consistency and systematicity of AI regulation while sensitively considering the context of the financial sector. Second, it substantively implements a risk-based approach, applying differentiated regulations commensurate with the risk level of AI. Strict regulatory requirements are imposed on high-risk areas that can significantly impact consumer rights, such as credit assessment and insurance premium calculation. Meanwhile, more flexible regulatory standards are applied to low-risk areas like fraud detection, thereby simultaneously enhancing the effectiveness and efficiency of regulations. Third, it attempts a comprehensive and coherent regulatory response to the complex risks posed by financial AI by strengthening the interconnectedness between AI system regulation, operational resilience regulation, and data regulation. Detailed regulations aimed at ensuring the transparency and explainability of AI systems, enhancing cybersecurity and operational stability, and creating a safe data utilization foundation operate in a complementary manner, maximizing the synergy effect of regulations. Fourth, it reinforces multi-layered safeguards for financial consumer protection. Individual legal systems, such as the right to demand explanations for AI decisions under the AI Act, the assurance of financial service operational stability through DORA, and the strengthening of data sovereignty based on FIDA, contribute to consumer protection. Simultaneously, through close interaction among these systems, a comprehensive prevention and relief system is established for new types of consumer damage caused by AI. Fifth, it seeks to effectively enforce the multi-layered regulatory system by establishing a supervisory system combining expertise and integration centered on the European Supervisory Authorities (ESAs). The ESAs conduct comprehensive and proactive supervision of the entire financial AI ecosystem in close cooperation with individual national supervisory authorities, thereby enhancing the coherence and reliability of the overall regulatory system. The EU's pioneering regulatory case provides important reference points for shaping the direction of domestic financial AI regulatory system development. Above all, the harmonious linkage between general AI laws and individual financial sector regulations, the establishment of regulatory standards that align with risk characteristics, the balanced design of data utilization and protection, the strengthening of financial consumer protection, and the establishment of an effective supervisory system are raised as urgent policy tasks. Furthermore, financial AI regulations should be flexible and executable to adapt to the rapidly changing technological environment while ensuring compatibility with global standards. Keywords