South Korea's AI Basic Act

The law's full name is the Framework Act on the Development of Artificial Intelligence and Establishment of Trust. The National Assembly passed it on December 26, 2024, merging many earlier bills, and it was promulgated in January 2025 with an effective date of January 22, 2026. The Ministry of Science and ICT (MSIT) is the lead agency, and many details, such as thresholds and exact labeling methods, are set by presidential enforcement decree and ministry guidance. The Act has two halves. The promotion half requires a national AI basic plan every three years, supports AI data centers, training data, startups and talent, and gives a legal footing to bodies such as a national AI committee and an AI safety institute. The trust half defines 'high-impact AI' as systems that may significantly affect human life, physical safety or fundamental rights in listed areas, including energy supply, drinking water, healthcare and medical devices, nuclear facilities, biometric analysis in criminal investigations, hiring and credit decisions, transport, key public services and student evaluation. Operators of such systems must implement risk management, explanation measures, user protection and human oversight, and keep documentation. For generative AI, businesses must notify users in advance that a product runs on generative AI and label outputs as AI-generated; content that is hard to distinguish from reality, such as deepfakes, must be clearly indicated. Foreign businesses above set thresholds must designate a domestic representative. A frequent misconception is that it copies the EU AI Act. It borrows risk-based ideas but has no list of banned practices, and its fines are modest: administrative fines capped at 30 million won for violations such as failing to label or appoint a representative. The government has also signaled a grace period before fines are actively imposed.