Financial Supervisory Service to Implement AI Risk Management in the Financial Sector

Introduction of the "AI Risk Management Framework (AI RMF) for the Financial Sector" in the First Quarter

The Financial Supervisory Service announced on January 15 that, in response to the rapid adoption of artificial intelligence (AI) in the financial sector, it will introduce the "AI Risk Management Framework (AI RMF) for the Financial Sector" within the first quarter of this year to enable financial institutions to independently control and manage risks.

The financial authorities have been supporting the financial industry from various angles to lead the AI transformation and drive financial innovation. However, concerns have grown that new types of risks-stemming from the complexity, opacity, and data dependency of AI-could infringe on the rights and interests of financial consumers or undermine financial stability.

In response, the Financial Supervisory Service plans to establish the AI RMF to provide a structural foundation that enables financial institutions to manage risks independently, ensuring that AI is utilized in a sound manner with a balance between innovation and responsibility.

The AI RMF will propose three main core processes: "governance," "risk assessment," and "risk control." Regarding governance, financial institutions will be required to establish decision-making bodies and dedicated teams for AI risk management, as well as develop relevant internal regulations to build a governance system. It is necessary to assess and manage potential risks at every stage-including planning, development, and operation-of AI systems under a framework of clear accountability and checks and balances.

For AI risk assessment, the key is for financial institutions to establish a comprehensive evaluation system based on a risk-based approach in order to systematically identify, measure, and manage AI risks. To control AI risks, financial institutions must implement differentiated controls and management according to the level of risk, and for ultra-high-risk AI, they must carry out all necessary procedures for risk control, such as re-examining whether to launch the AI system.

However, the Financial Supervisory Service explained that the AI RMF will be a voluntary guideline without legal enforceability, allowing each company to strengthen, relax, or selectively apply it according to their own circumstances.

A representative from the Financial Supervisory Service said, "We plan to distribute the draft AI RMF to the financial industry through associations for each business sector and gather sufficient feedback through explanatory sessions, before finalizing and implementing the framework within the first quarter of this year."