South Korea AI Safety Institute

The AI Safety Institute (AISI) of South Korea was officially launched on November 27, 2024, at the Pangyo Global R&D Center in Seongnam-si, Gyeonggi Province. It was established as a subsidiary organization under the Electronics and Telecommunications Research Institute (ETRI) and operates under the oversight of the Ministry of Science and ICT (MSIT). The institute was created in direct response to the AI Seoul Summit held in May 2024, where South Korea joined nine other nations and the European Union in committing to a network of international AI safety institutes. The institute is directed by Kim Myung-joo (Kim Myuhng-Joo), an information security professor at Seoul Women's University who brings over 30 years of expertise in security and digital ethics. She previously founded the Right AI with Security and Ethics Research Center (RAISE) in 2019 and introduced Korea's first AI ethics framework, the Seoul PACT (emphasizing Publicness, Accountability, Controllability, and Transparency). Her three-year term as director runs from November 12, 2024 to November 11, 2027. At launch, the institute had 12 staff members consisting of the director and personnel seconded from ETRI and the Telecommunications Technology Association (TTA), with plans to grow to approximately 30 employees by end of 2025. The institute's annual budget is approximately 12 billion Korean won, comprising around 8 billion KRW for R&D (solution development) and 4 billion KRW for non-R&D operations including safety assessment indicators. The institute organized a 24-member Korea AI Safety Consortium at launch, including 13 industry partners (Naver, KT, Kakao, LG AI Research, SKT, Samsung Electronics, and others), 6 academic institutions (Seoul National University, KAIST, Korea University, Sungkyunkwan University, Soongsil University, and Yonsei University), and 5 research institutes (TTA, NIA, KISDI, IITP, and SPRi). The consortium discusses AI safety evaluation frameworks, upgrades global safety reports for Korean contexts, and facilitates joint research and expert exchanges. On the international front, Korea AISI is a member of the International Network of AI Safety Institutes comprising ten countries. Its international cooperation includes joint research on deepfakes, multilingual and multicultural AI testing, cybersecurity testing, and agentic AI safety. The institute has also collaborated with the U.S. State Department, including a workshop on AI security at Lawrence Livermore National Laboratory. The institute explicitly positions itself as a research and support hub rather than a regulatory body, aiming to help Korean AI companies minimize risk factors that could hinder their global competitiveness while advancing national AI safety governance.