OpenAI is launching a "Korea Cyber Action Plan" to expand access to its latest artificial intelligence (AI) cybersecurity models to the government and major corporations in Korea to strengthen domestic cyber defense capabilities. The plan is to selectively open high-performance AI models to trusted institutions and corporations to preemptively respond to security threats.
Jason Kwon, OpenAI chief strategy officer (CSO), said at a press briefing at the JW Marriott Hotel in Seocho District, Seoul, on the 27th, "The latest cyber AI capabilities should not remain with a small group, and Korea's key defense actors must be able to use them to strengthen collective security and public safety," adding, "Through the Korea Cyber Action Plan, OpenAI will work to help strengthen Korea's capabilities in the field of cybersecurity."
The Korea Cyber Action Plan is a program that supports strengthening AI-based cyber defense capabilities for the government, public institutions, and domestic corporations, and it operates under OpenAI's cybersecurity initiative "Daybreak."
The program consists of "GTAC," a trust-based access program for government agencies, and "TAC," a program for major corporations responsible for core national industries. Participating institutions can secure access to OpenAI's latest high-performance AI models, including GPT-5.5 Cyber, to detect emerging cyber threats and address vulnerabilities. According to OpenAI, Korea's participation in the GTAC program is the third in the world after the United States and Canada.
Kwon said, "Given the rapid pace of AI development, it is important that trusted defense actors secure relevant capabilities ahead of malicious actors," adding, "The key is to help swiftly find and respond to vulnerabilities."
One cited reason behind this policy expansion is growing concern that high-performance AI models could be misused for hacking and other purposes. Anthropic, OpenAI's competitor, is also running "Project Glasswing," which provides its vulnerability-detection–focused AI model "Claude Mythos Preview" to key institutions on a limited basis to verify and share security system vulnerabilities.
OpenAI said it has been in contact with key institutions for this collaboration. On the 18th, Sasha Baker, OpenAI head of national security policy, visited the Ministry of Science and ICT, the Financial Services Commission, and the Korea Internet & Security Agency (KISA), and conducted demonstrations of cyber-specialized models. The day before, Kwon held a meeting with Vice Minister Ryu Je-myeong of the Ministry of Science and ICT and others to discuss cooperation plans.
Meanwhile, OpenAI also emphasized that ChatGPT usage is rapidly expanding in the Korean market. According to OpenAI, Korea ranked in the global top 10 for weekly active users (WAU) of ChatGPT, corporate customers, and paid subscribers. In particular, weekly active users in Korea of the coding AI tool "ChatGPT Codex" were tallied as having increased tenfold from the beginning of the year.