A New Era of AI Legislation

South Korea is poised to enforce its comprehensive Act on the Development of Artificial Intelligence and Establishment of Trust (AI Basic Act) on January 22, 2026. This legislation marks the country as the second jurisdiction globally, following the EU, to implement a horizontal AI law. The Act serves as the cornerstone of the Lee Jae Myung administration's ambitious vision, announced in June 2025, to position South Korea among the world’s top three AI powerhouses. While a proposed three-year grace period for full regulatory enforcement has been discussed, the law formally establishes the legal infrastructure for both industrial promotion and safety compliance.

The Risk-Based Classification System

The governance framework adopts a nuanced, risk-based approach that categorises AI systems into distinct tiers with specific compliance obligations:

• High-Impact AI: Systems affecting human life, safety, or fundamental rights, such as those used in nuclear power, traffic control, recruitment, and loan evaluation, require operators to conduct pre-deployment assessments, notify users, and implement rigorous safety frameworks.
  

• Generative AI and Deepfakes: To combat misinformation, operators of systems that generate text, audio, or video must provide advance notification to users and clearly label AI-created products. Deepfakes specifically require clear indicators to prevent them from being mistaken for reality.
  

• High-Performance AI: Defined by a cumulative compute threshold used during training, these systems mandate lifecycle risk identification and the establishment of incident response management systems.
  

Institutional Oversight and Enforcement

A federated network of regulatory bodies is tasked with enforcing these new standards:

• Ministry of Science and ICT (MSIT): Acts as the primary enforcer, authorised to issue correction orders and impose administrative fines of up to KRW 30 million (~USD 21,700) for violations of safety or labelling standards.
  

• Personal Information Protection Commission (PIPC): Spearheads reforms to the Personal Information Protection Act (PIPA) to facilitate the safe use of pseudonymized data for AI training and manage privacy risks.
  

• Korea Fair Trade Commission (KFTC): Monitors infrastructure monopolisation and collaborates with the Korea Consumer Agency to crack down on "AI washing" (deceptive advertising).
  

• Sector-Specific Regulators: Agencies like the Ministry of Food and Drug Safety (MFDS) and the Financial Services Commission (FSC) have issued specialised guidelines for medical devices and financial AI, respectively.
  

Conclusion

As the January 22, 2026, enforcement date approaches, South Korea's regulatory landscape has shifted from voluntary guidelines to binding statutory obligations. By balancing the protection of human dignity with aggressive industrial support, including resolving copyright disputes over text and data mining, the AI Basic Act establishes a robust foundation for South Korea's ascent as a global leader in safe and trustworthy artificial intelligence.