この記事は現在英語でのみご利用いただけます。
South Korea's AI Framework Act 2026: The First Comprehensive AI Law in Asia-Pacific
South Korea's AI Framework Act took effect January 22, 2026 — making it the first country in APAC to have a comprehensive AI law with real obligations for high-impact AI systems. Unlike Japan's approach, South Korea's law requires transparency, risk assessments, human oversight, and carries financial penalties.
Key Takeaways
South Korea's AI Framework Act (enacted January 2025, effective January 22, 2026) is the first comprehensive AI law in the Asia-Pacific region with mandatory compliance obligations for high-impact AI.
High-impact AI is defined as AI used in healthcare, energy, water, nuclear, biometrics, employment, credit, transport safety, public administration, and education — sectors with potential to significantly affect human life or fundamental rights.
Operators of high-impact AI must conduct impact assessments, implement risk management, ensure explainability, provide human oversight mechanisms, and maintain documentation.
Generative AI operators must label AI-generated content to inform users — a transparency requirement that aligns with EU AI Act Article 50 requirements.
The Act applies to foreign entities whose AI systems affect Korean users. Foreign AI companies must designate a Korean domestic representative if they meet the criteria in the Enforcement Decree (effective January 22, 2026).
"情報提供のみを目的としています。この記事は法律、規制、財務または専門的なアドバイスを構成するものではありません。具体的なアドバイスについては、資格を持つ専門家にご相談ください。"
Why South Korea's AI Framework Act matters globally
South Korea's Basic Act on the Development of Artificial Intelligence and Establishment of a Foundation for Trustworthiness — the AI Framework Act — entered into force on January 22, 2026. It is the first comprehensive AI law in the Asia-Pacific region with mandatory compliance obligations backed by financial consequences. Its approach sits between Japan's innovation-first soft law and the EU's risk-based hard regulation.
The Act was passed by South Korea's National Assembly on December 26, 2024, promulgated on January 21, 2025, following the consolidation of 19 separate AI bills into a single unified framework. The Presidential Enforcement Decree, enacted January 21, 2026, provides operational details on key obligations including the computation threshold for high-impact AI operators and domestic representative requirements for foreign entities.
The high-impact AI framework
The AI Framework Act's primary compliance burden falls on operators of "high-impact AI" — AI systems that may significantly affect human life, safety, or fundamental rights in specific sectors. The Enforcement Decree specifies these sectors: healthcare and medical devices; energy supply (electricity, gas); water production and supply; nuclear energy; biometric investigations; employment and HR (including hiring decisions); credit decisions; transportation safety; public administration; and education.
For AI systems in these categories, operators must: conduct AI impact assessments before deployment; implement risk management systems appropriate to the impact level; provide users with meaningful explanations of AI-influenced decisions; ensure adequate human oversight mechanisms; and maintain documentation demonstrating compliance. These requirements have substantive content — they are not disclosure-only obligations.
Generative AI transparency requirements
Operators of generative AI systems that produce content must label outputs to inform users that the content was AI-generated. This applies to text, images, audio, and video generation. The transparency requirement aligns with EU AI Act Article 50's AI-generated content disclosure obligations, though the specific technical implementation requirements differ.
Extraterritorial scope: foreign AI companies must comply
The AI Framework Act applies to foreign entities whose AI systems affect Korean users or markets. This extraterritorial reach means that global AI companies — including those headquartered in the US, Europe, China, and Japan — must assess whether their Korean-facing activities bring them within scope. The Enforcement Decree specifies criteria for the domestic representative requirement: foreign AI companies meeting defined threshold criteria must designate a Korean representative to liaise with MSIT (Ministry of Science and ICT) on compliance matters.
What organisations need to do
Assess whether your AI systems qualify as "high-impact" under the Enforcement Decree's sector definitions. If you operate generative AI systems accessible by Korean users, implement content labelling. If you are a foreign entity with Korean-facing AI, determine whether you trigger the domestic representative requirement. Build impact assessment capabilities and risk management documentation for high-impact systems. The Ministry of Science and ICT (MSIT) is the primary regulatory authority for the Act — engage with MSIT guidance as it develops through 2026.