OpenAI will open its latest high-performance cyber-specialized artificial intelligence (AI) models to the Korean government, public agencies and companies. Amid growing cybersecurity threats from AI advances, the move is intended to help Korea's key defense actors strengthen their cyberattack response capabilities by leveraging OpenAI's latest models.
OpenAI announced the "Korea Cyber Action Plan" containing these details at a press conference held Thursday at the JW Marriott Hotel in Seocho-gu, Seoul. The plan was prepared as part of "Daybreak," OpenAI's cybersecurity initiative.
At the core of the action plan is Korea's participation in the "Government Trusted Access Collaboration (GTAC)" program. GTAC is a program through which OpenAI grants access to its latest high-performance AI models to vetted governments and public agencies. Korea, alongside Japan, has become the third country in the world to join GTAC, following the United States and Canada.
"Cutting-edge cyber AI capabilities should not remain in the hands of a few; Korea's key defense actors should be able to use them to strengthen collective security and public safety," said Jason Kwon, OpenAI's Chief Strategy Officer (CSO). He added, "Through the Korea Cyber Action Plan, OpenAI will work closely with the government, public agencies and companies to support the strengthening of Korea's cybersecurity capabilities."
Ahead of the announcement, OpenAI had been laying the groundwork for cooperation with the Korean government. On the 18th of this month, Sasha Baker, head of national security policy at OpenAI, visited Korea and demonstrated the latest cyber-specialized models to major government ministries and public agencies, including the Ministry of Science and ICT, the Ministry of Foreign Affairs, the Ministry of the Interior and Safety, the Financial Services Commission, the National AI Strategy Committee and the Korea Internet & Security Agency (KISA). CSO Kwon also met with Ryu Je-myung, Second Vice Minister of Science and ICT, and others on the 26th to discuss cooperation in cybersecurity.
Following the public sector, OpenAI will also expand cooperation with private companies. Separately from GTAC, OpenAI plans to pursue the participation of major Korean conglomerates in TAC, a program for private companies. The company is currently in discussions with Korean firms that wish to join TAC. To participate in TAC, private companies must obtain approval from OpenAI.
OpenAI plans to strictly limit access to TAC. Access will be granted only to white-hat hackers, defense researchers and specialized security organizations that have undergone thorough identity verification and qualification screening. "Cybersecurity must go beyond post-incident response and be embedded proactively during the development process," CSO Kwon said. "We are running this program so that trusted defense actors can quickly identify vulnerabilities, remediate and patch them, and block cyberattacks."
OpenAI's move is also seen as a strategy to expand its public and private customer base in competition with rival Anthropic. Anthropic operates its security alliance "Project Glasswing" with a select group of companies and institutions. The Ministry of Science and ICT is reportedly continuing discussions with Anthropic to join Glasswing.
Behind OpenAI's expanded cooperation with Korea is the country's high AI adoption rate. According to OpenAI, Korea ranks among the world's top 10 in ChatGPT weekly active users, enterprise customers and paid subscribers. Korea is also among the top five countries in ChatGPT Codex usage and engagement.
"Korean users have already embraced AI, are using it, and are beginning to integrate it into their lifestyles and ways of working," CSO Kwon said. "This is one of the reasons OpenAI seeks to cooperate more deeply in Korea's AI transformation."
