대출 알고리즘 차별로부터의 금융소비자 보호를 위한 법적 개선 방안

As generative AI begins to be applied in the financial sector, financial institutions are rapidly analyzing vast amounts of data to inform credit assessments and personalized product recommendations. While such developments can bring greater efficiency to decision-making, concerns have been raised that if training data contains bias, its effects may extend to the transaction conditions offered to financial consumers. This study examines regulatory improvement measures from a financial consumer protection perspective regarding algorithmic discrimination in generative AI-based financial transactions. To prevent algorithmic discrimination, it is first necessary to clearly define the prohibited grounds of discrimination and ensure consistency among overlapping regulatory frameworks. Since a wide variety of data types are utilized in applying generative AI to finance, careful consideration must be given to balancing technological application with the protection of data subjects' rights. As automated decision-making becomes more widespread, data subjects must be able to request sufficient explanation regarding the decision-making process, and mechanisms must be established to identify potential bias in advance through impact assessments. Where data subjects refuse the use of their data or the application of automated decision-making, alternative assessment methods must be prepared to protect them from any resulting disadvantage. This study also proposes a comparative disclosure method whereby transaction conditions produced through algorithm-based decision-making and those calculated through traditional methods are presented side by side, assisting consumers in assessing whether algorithmic discrimination has occurred. The right to demand explanations can function as regulatory checks, contributing to fairness, transparency, and the correction of discriminatory bias. Given that discriminatory outcomes in algorithmic decision-making are structurally difficult to resolve through individual rights remedies, this study proposes a broader interpretation of the Financial Consumer Protection Act so that the Financial Services Commission can function as an active regulator removing structural discrimination factors, beyond simply serving as a reactive dispute mediator. It is hoped that the regulatory measures proposed in this study will contribute to correcting bias in financial consumer-related data, ensuring that financial consumers receive appropriate services, and supporting the realization of a sustainable financial inclusion policy in the era of generative AI.