Korea's AI Framework Act takes effect in January 2026, creating obligations for high-impact and generative AI systems.
Korea's AI Basic Act, passed in December 2024 and entering force in 2025, establishes the world's first enacted national AI statute outside of the EU. Implementation guidance and ministerial ordinances — covering high-impact AI classification, transparency requirements, and the operational protocols for the AI Safety Institute Korea — are expected to take effect through 2026.
Key provisions now in effect include: mandatory transparency disclosure to end-users when interacting with AI systems, a public register of government-deployed AI, and sector-specific rules for AI in hiring, credit scoring, and healthcare that require algorithmic impact assessments. The AI Safety Institute Korea (AISK) functions as the national testing and evaluation body — analogous to the UK AI Safety Institute — and has begun collaborative model evaluations with NIST and the EU AI Office.
For APAC enterprises deploying AI systems to Korean end-users or operating Korean subsidiaries, the Act's transparency and impact-assessment obligations are immediately actionable. AIMenta Korea advises on the interaction between the AI Basic Act and Korea's Personal Information Protection Act (PIPA), which creates a complex dual-compliance environment for AI systems that process personal data — the majority of commercial AI applications.