Korea Becomes Third Country Globally to Join OpenAI GTAC... Partners for AI Security (Comprehensive)

Participation alongside Japan after the US and Canada

Ryu Jaemyung: "Secured access to OpenAI's latest models"

The South Korean government is launching comprehensive cooperation with global AI company OpenAI to address the rapidly evolving security threats posed by artificial intelligence (AI). The Ministry of Science and ICT has formalized its participation in OpenAI's Government & Trust Agency Collaboration (GTAC) program, which is designed for governments and institutions, and has secured access to OpenAI's latest high-performance models. Following the United States and Canada, South Korea, together with Japan, has become the third country globally to join GTAC.

The Ministry of Science and ICT announced on the 27th that it held a meeting with OpenAI the previous day to discuss measures for responding to security threats, securing AI safety and trust, and exploring future cooperation. During the meeting, Ryoo Jemyung, Second Vice Minister of the Ministry of Science and ICT, met with Jason Kwon, Chief Strategy Officer (CSO) of OpenAI, and other representatives, continuing in-depth discussions on various AI security threats.

The Trust Agency Collaboration (TAC) operated by OpenAI is a trust-based cyber access program that selectively provides the latest AI models to trusted partners, and GTAC is its government and institutional version. Organizations participating in the GTAC program gain access to OpenAI's latest high-performance AI models, enabling them to respond proactively to emerging cyber threats. This helps prepare for and develop response strategies against the potential misuse of rapidly advancing AI models in cyberattacks.

South Korea's participation in GTAC, alongside Japan, marks the first such case in Asia. The Korea Internet & Security Agency (KISA) will serve as the implementing body. Both parties agreed to continue active discussions on the use of AI models in the field of cybersecurity, starting with this participation in GTAC.

At the press conference held that day, Jason Kwon, CSO of OpenAI, stated, "Cutting-edge cyber AI capabilities should not be concentrated in the hands of a few. Key defense entities in Korea should be able to leverage these capabilities to enhance collective security and public safety. We will cooperate with the government, public institutions, and businesses to strengthen Korea's capabilities in the cyber security sector."

OpenAI is also launching the "Korea Cyber Action Plan," which supports high-performance AI cyber models. The plan aims to go beyond existing cybersecurity response methods that focus on identifying and patching vulnerabilities, and instead support the government, public institutions, and domestic companies in strengthening cyber defense capabilities from the software design stage.

Ryoo Jemyung, Second Vice Minister of the Ministry of Science and ICT (right), is taking a commemorative photo with Jason Kwon, Chief Strategy Officer (CSO) of OpenAI, at the ThreeIFC conference room in Yeongdeungpo-gu, Seoul on the afternoon of the 26th. Provided by the Ministry of Science and ICT.

Additionally, both parties discussed cooperation measures to ensure AI safety. The Ministry of Science and ICT recently requested the establishment of a cooperative relationship between the AI Safety Institute and OpenAI to pursue practical collaboration, such as AI safety assessments and joint research, in response to various AI risks that have emerged due to improvements in AI performance and an expanded range of applications. OpenAI also agreed to actively review this proposal.

Vice Minister Ryoo emphasized, "Through our collaboration with OpenAI, South Korea has established a foundation to preemptively respond to AI security threats. The Ministry of Science and ICT will continue to actively engage in cooperation and practical discussions with global AI companies to enhance domestic AI security capabilities."

Meanwhile, the government is also continuously seeking participation in "Project Glasswing," a global AI security consortium led by Anthropic. Although Anthropic has established a closed cooperation system with only 52 companies and institutions, including Google, Amazon, and Microsoft (MS), the government and private sector are exploring various ways to participate. However, as of now, no South Korean company or institution has been included in the project.

A representative from the Ministry of Science and ICT said, "With growing concerns over security threats posed by Anthropic's high-performance AI model 'Mythos,' we are continuously seeking participation in Project Glasswing. In addition to our collaboration with OpenAI, we will expand cooperation with other global AI companies to prepare for evolving security threats."

This content was produced with the assistance of AI translation services.