The Ministry of Science and ICT announced on the 27th that it has confirmed its participation in the U.S. company OpenAI's Government and Agency Trust-based Access Program (GTAC). With Samsung Electronics also exploring cooperation with Anthropic's AI security project 'Glasswing' in the U.S., moves to strengthen AI-based cybersecurity responses are accelerating in both the public and private sectors.
The Ministry of Science and ICT (MSIT) formalized its participation in GTAC yesterday during a meeting attended by Ryu Je-myung, second vice minister, and Jason Kwon, chief strategy officer (CSO) of OpenAI, to seek cooperation plans such as responding to AI security threats and ensuring AI safety and trust.
The Korea Internet & Security Agency (KISA) will handle the practical working-level tasks to secure access rights to OpenAI's latest high-performance AI models, and plans to utilize them for cybersecurity purposes, such as analyzing vulnerabilities in the security-specialized model 'GPT-5.5-Cyber'.
As cybersecurity concerns grew after Anthropic unveiled its security-specialized AI model 'Claude Mythos Preview' last month, the government has been exploring participation in global AI security cooperation channels from a national security perspective. While South Korea has not yet participated in Anthropic-led Project Glasswing, it has listed its name alongside Japan as the first countries in Asia to participate in OpenAI's GTAC.
The private sector, including Samsung Electronics, is also seeking security cooperation plans with global AI companies on a company-by-company basis. This is because as U.S. AI companies form security alliances centered on their own big tech companies, the necessity for Korean companies to improve the security verification capabilities of their own products and services has grown.
A high-ranking industry official said, “Samsung Electronics is exploring cooperation with Anthropic's Project Glasswing for multiple solutions that require security diagnosis, including the Galaxy security platform 'Knox'.”
Knox is a security platform embedded in Galaxy devices. It is interpreted that the company pushed for participation to identify potential vulnerabilities at an early stage across the operating system (OS), firmware, and security platform of Galaxy smartphones.
The domestic security industry is also seeking solidarity measures for AI security responses. Apart from individual companies participating in global AI projects, discussions are underway to establish a cooperative body involving security firms, public agencies, demanding companies, and AI companies.
Based on this, measures are being discussed to apply usable AI models to corporate security inspections to enhance verification capabilities and to link them to cyber responses in the public sector. Supporting the development of security-specialized AI models tailored to the domestic security environment is also cited as a major task.
Hong Joon-ho, professor of convergence security engineering at Sungshin Women's University, said, “With the proliferation of generative AI and AI agents, security has emerged as a key variable determining industrial competitiveness and sovereignty,” adding, “The government and public agencies must share global cooperation experiences with domestic companies and support them to build technological capabilities that can be utilized in actual security fields.”
