- The Framework Act on AI Development and Trust Foundation, also known as the AI Basic Act, will officially take effect on January 22, 2026, in South Korea.
- The revised law strengthens monitoring of generative AI content while expanding support for research, startups, and digital inclusion.
- New regulations mandate that commercial AI service providers must disclose AI-generated content through notifications or technical measures like watermarking.
- The goal is to curb deepfakes, misinformation, and misleading advertising.
- The law defines “high-impact AI” for the first time, applying to systems trained with at least 100 septillion floating-point operations.
- These systems will be subject to strict safety requirements, risk management plans, social impact monitoring, and direct inspections by authorities.
- The Presidential Council on National AI Strategy is granted legal status, becoming the central hub for coordinating and overseeing AI policy.
- The government is empowered to require businesses to submit data and to conduct on-site inspections.
- The law allows universities and private enterprises to establish AI research institutes, not limiting this to the Ministry of Science.
- The public sector is required to prioritize the procurement and use of AI products and services to stimulate market demand.
- Civil servants implementing AI will be immune from personal liability absent intent or gross negligence.
- The law emphasizes AI accessibility for the disabled, the elderly, and vulnerable groups, accompanied by subsidy mechanisms.
- Businesses are concerned that the definition of high-impact AI remains vague, potentially placing a heavy burden on startups.
- The Startup Alliance warns that common tools like machine translation or photo editing could also be subject to watermarking requirements.
- The government is applying a grace period of at least one year, prioritizing guidance over punishment.
📌 The Framework Act on AI Development and Trust Foundation, also known as the AI Basic Act, will officially take effect on January 22, 2026, in South Korea. The Presidential Council on National AI Strategy is granted legal status, becoming the central hub for coordinating and overseeing AI policy. The government is empowered to require businesses to submit data and to conduct on-site inspections. The public sector is required to prioritize the procurement and use of AI products and services to stimulate market demand. Civil servants implementing AI will be immune from personal liability absent intent or gross negligence. The government is applying a grace period of at least one year, prioritizing guidance over punishment.
