South Korea eases AI rules as oversight tightens

Regulators will accelerate network separation reforms for financial AI systems.

South Korea’s Financial Services Commission (FSC) introduced an updated artificial intelligence (AI) guidelines designed to manage risks whilst easing technical barriers.

FSC Vice Chairman Kwon Dae-young outlined a new regulatory framework focused on three core areas: maintaining a level playing field without AI-specific advantages, establishing clear data accountability, and managing unpredictable cybersecurity threats.

To support this transition, regulators plan to fast-track the easing of network separation requirements for financial AI systems. 

The government will also update rules regarding personal credit data consent and pseudonymisation—the process of replacing private data with artificial identifiers to protect user identity.

The updated AI guidelines, which took effect on 22 June, establish seven self-regulatory principles for all financial and fintech firms. 

These cover corporate governance, legal compliance, credibility, financial stability, consumer protection, and security. 

Under these rules, AI must function strictly as an assistant, leaving ultimate decision-making responsibility with human supervisors.

High-impact AI applications must comply with the sterner requirements of the Framework Act on the Development of Artificial Intelligence.

Firms are advised to adopt these principles based on their specific resources and risk profiles. 

To assist with the transition, the Financial Supervisory Service and the Financial Security Institute will launch an AI risk management framework, a dedicated security guidebook, and a support helpdesk.

The FSC will use feedback from the meeting to identify necessary regulatory reforms and intends to launch controlled testing of AI agents under its financial regulatory sandbox program during the second half of this year.