The Framework Act on the Development of AI and Establishment of a Foundation for Trustworthiness is South Korea's horizontal AI statute. The National Assembly of the Republic of Korea passed it on 26 December 2024, it was promulgated on 21 January 2025, and it took effect on 22 January 2026. The lead ministry is MSIT. This page summarizes the statute as described across the secondary sources listed below; the primary text is hosted at the official Korean government portal aibasicact.kr. The legislation page is South Korea AI Basic Act — Source Summary.
According to these sources, it is the first comprehensive AI framework law in Asia and the second globally after the EU AI Act (EU AI Act (Regulation 2024/1689)). It combines risk-tier regulation with pro-innovation governance, statutory footing for an AI Safety Institute, and mandatory generative-AI labelling.
Structure and scope
The Act establishes a "high-impact AI" tier covering sectors with significant rights or safety impact, and an "AI of national significance" tier added late in the legislative process, analogous to the EU's general-purpose AI with systemic risk. It mandates generative-AI labelling under Article 31 and asserts extraterritorial reach, with a domestic-representative requirement for foreign operators.
The high-impact sectors are energy; healthcare and medical devices; nuclear safety; biometric information; recruitment; credit and financial services; public services with significant rights impact; transportation and autonomous driving; and criminal justice.
Obligations
Core obligations for high-impact AI comprise a risk management system; impact assessments covering human rights and safety; explanation to affected users; human oversight capable of intervention; and a user-protection programme.
Generative-AI transparency requirements consist of mandatory labelling of outputs that mimic real-world content, disclosure of interaction, and watermarking or technical equivalents.
Governance
Governance is structured around a National AI Committee chaired by the President, with MSIT as lead regulator and the Korea AI Safety Institute (K-AISI) holding statutory footing under the Act. Sectoral regulators retain their existing sectoral authority. The statutory footing of K-AISI distinguishes it from the AI Safety Institutes of the United Kingdom and the United States, including the UK AI Safety Institute (AI Security Institute), which does not have an equivalent statutory basis.
Enforcement
Enforcement provides for administrative fines of up to KRW 30 million per infraction and corrective orders. The statute itself contains no criminal penalties. The sources note that the per-infraction KRW 30 million penalty ceiling is low by international comparison, and that implementing regulations may add a turnover-based uplift, though this was unresolved at commencement. The threshold specifics for "AI of national significance" — compute, parameters, and capabilities — were deferred to implementing regulation.
Key claims
- High confidence: The National Assembly passed the Act on 26 December 2024; promulgation 21 January 2025; effective 22 January 2026.
- High confidence: Korea is the second jurisdiction globally (after the EU) with a comprehensive horizontal AI statute, and the first in Asia.
- High confidence: Generative-AI labelling is mandatory for outputs mimicking real-world content.
- High confidence: The Korea AI Safety Institute has statutory footing under the Act, distinguishing it from the UK and US AISIs.
- High confidence: Extraterritorial reach and the domestic-representative requirement mirror GDPR and PIPA.
- Medium confidence: The per-infraction KRW 30 million penalty ceiling is low by international comparison; implementing regulations may add a turnover-based uplift, but this was unresolved at commencement.
- Medium confidence: The "AI of national significance" threshold specifics (compute, parameters, capabilities) were deferred to implementing regulation.
Relation to other pages
The Act is the first South Korean legislation covered here, extending coverage of Asian AI regulation beyond China and Singapore. The sources describe the EU AI Act as the principal model, characterizing the Korean approach as an adaptation rather than a clone. Within Asia, it contrasts structurally with Japan AI Promotion Act — Source Summary, setting a risk-tier-plus-rights design against a pro-innovation, soft-law approach. The generative-AI labelling mandate connects to AI and Content Moderation, and the regional-template framing connects to AI Sovereignty.
No direct contradictions with existing wiki pages were identified. The source material contains no PII, credentials, or sensitive identifiers.
Relationships
- supports: AI Basic Act (legislation page)
- related: EU AI Act (Regulation 2024/1689) — principal model
- related: Japan AI Promotion Act — Source Summary, Brazil AI Bill (PL 2338/2023) — Source Summary, Canada AIDA (Bill C-27, Part 3) — Source Summary, Australia Voluntary AI Safety Standard — Source Summary — the international-completeness cohort
- related: China — Interim Measures for the Management of Generative AI Services, China — Provisions on the Administration of Deep Synthesis Internet Information Services — Asian regulatory contrast
- related: UK AI Safety Institute (AI Security Institute), AI Sovereignty, AI Safety Cases and Frameworks
Sources
- aibasicact.kr (official Korean government portal):
https://aibasicact.kr/ - Future of Privacy Forum, "South Korea's New AI Framework Act: A Balancing Act Between Innovation and Regulation"
- IAPP, "Analyzing South Korea's Framework Act on the Development of AI"
- US Department of Commerce / trade.gov, "South Korea Artificial Intelligence (AI) Basic Act"
- Cloud Security Alliance, "What You Need to Know About South Korea's AI Basic Act" (Mar 2025)
- TLP Advisors, "The Closing Act of 2024: South Korea's AI Basic Act" (Jan 2025)
- Securiti, "An Overview of South Korea's Basic Act on AI"