Republic of Korea: Ministry of Science and Information and Communication Technology issued draft Enforcement Decree of Framework Act on Development of Artificial Intelligence and Establishment of Trust Foundation including design requirements

On 8 September 2025, the Ministry of Science and Information and Communication Technology (MSICT) released the draft Enforcement Decree of the Framework Act on the Development of Artificial Intelligence and Creation of a Trust Foundation (AI Framework Act). The draft outlines design requirements, including obligations for prior notification to users of generative and high-impact artificial intelligence (AI), display of results through watermarking, recognition of invisible watermarks, and mandatory labelling of deepfake outputs taking into account user age and physical condition, with exemptions for internal business use or clearly identifiable generative results. The draft required safety assurance for high-performance AI with a cumulative learning volume of 10^26 FLOPS or more, including risk identification, assessment, mitigation, and emergency response planning. High-impact AI, defined as systems significantly affecting life, physical safety, or fundamental rights, including those in energy, healthcare, nuclear power, transportation, and education, would be subject to implementation plans for reliability, user protection, malfunction prevention, security measures, and feedback procedures. While the AI Framework Act imposes fines of up to KRW 30 million for violations, the draft Enforcement Decree introduced a guidance period for fines, with the duration to be determined after consultation with stakeholders. The MSICT announced that briefing sessions would be held with companies, organisations, civic groups, overseas operators, and content industries before the subordinate statutes are finalised and published.