The Ministry of Science and ICT said on the 17th that it signed a memorandum of understanding (MOU) at the ETRI Seoul office with the AI Safety Institute and OpenAI to strengthen cooperation on AI safety in high-risk fields.
The MOU aims to expand the AI cooperation framework established by the agreement between the ministry and OpenAI in Oct. last year to the field of AI safety.
Based on safety cooperation measures discussed during two meetings this year between the ministry's second vice minister and senior OpenAI officials, the two organizations began full-fledged, practical cooperation.
Through the agreement, the AI Safety Institute and OpenAI will share knowledge and best practices on safety assessment methodologies and benchmarks for each high-risk domain, and exchange technical information to develop an evaluation framework that reflects the Korean language and Korea's social context. They also agreed to continue cooperating to establish an AI safety evaluation system that can be applied internationally.
Korea is the fourth country, after the United States, the United Kingdom and Japan, with which OpenAI has signed an MOU with a national AI safety institute. The ministry said Korea will strengthen its role in the international cooperation network for validating risks and establishing evaluation standards for cutting-edge AI through this agreement.
Lee Jin-su, director general for AI policy planning at the ministry, said, "It is time to strengthen cooperation with global leading corporations and advance AI safety evaluation frameworks to ensure the safety of cutting-edge AI models such as high-performance AI and autonomous agent AI," adding, "We expect both organizations, through this agreement, to actively cooperate to secure the safety of cutting-edge AI in response to the rapidly changing global AI technology and usage environment."
Kim Myeong-ju, head of the AI Safety Institute, said, "As AI's impact on national core infrastructure and security grows, rigorous safety assessments for high-risk fields are becoming a necessity, not a choice," adding, "The AI Safety Institute will work with OpenAI to scientifically validate the risks of cutting-edge AI and help build an evaluation framework that is internationally accepted."
Lee Sang-hyeon, head of Asia-Pacific policy at OpenAI, said, "Korea is an important country where the use and innovation of AI technology are progressing rapidly," adding, "We are pleased that this cooperation will allow us to contribute together to developing trustworthy AI and fostering a safe usage environment."