OpenAI extends cyber program to Korea

Why it matters

OpenAI said on Wednesday it is extending its cybersecurity program to South Korea, opening access for government agencies, public institutions and companies to its latest defensive AI tools. The rollout includes Governmental Trusted Access for Cyber, or GTAC, and access to GPT-5.5-Cyber, a model the company says is tuned for specialized defensive security work. Jason Kwon, OpenAI’s chief strategy officer, announced the move at a press briefing in Seoul. South Korea joins the United States and Canada in the government-focused program and, alongside Japan, becomes one of the first countries in Asia to receive the access. ### What exactly is Korea getting access to? GTAC gives verified government agencies and public institutions access to OpenAI’s cybersecurity-focused models under a restricted-access framework. The Korea Times reported that Korean institutions will gain access to GPT-5.5-Cyber, while OpenAI said the model is being rolled out in limited preview for defenders securing critical infrastructure and other specialized workflows. (koreaherald.com) OpenAI said Trusted Access for Cyber is designed to lower refusals for approved defensive tasks such as vulnerability identification, malware analysis, reverse engineering, detection engineering and patch validation, while still blocking malicious activity including credential theft, stealth, persistence and exploitation of third-party systems. ### Who in South Korea will run the program? (koreatimes.co.kr) The Korea Internet & Security Agency, or KISA, will oversee the program in practice in South Korea, according to the Korea Times. The same report said OpenAI plans to expand trusted-access arrangements beyond government bodies to private companies tied to Korea’s key industries. The Ministry of Science and ICT said on May 27 that South Korea would participate in OpenAI’s government-and-institution trust-based access program. (openai.com) A day earlier, Kwon met Second Vice Minister Ryu Je-myung in Seoul to discuss AI security threats, safety and trust, according to Korean media reports. ### Why did OpenAI pick Korea now? Kwon said OpenAI chose Korea because of the country’s semiconductor industry, engineering talent, advanced digital infrastructure and rapid adoption of AI services. (koreatimes.co.kr) The Korea Herald reported that Kwon also said Korea ranks among OpenAI’s top 10 markets globally in weekly ChatGPT users, enterprise customers and paid subscribers. October was an earlier marker in the relationship. (koreaherald.com) The Korea Herald reported that OpenAI signed a memorandum of understanding with South Korea’s Ministry of Science and ICT in October, and held an AI security workshop earlier in May with Korean security, financial and diplomatic agencies. ### How does this fit with OpenAI’s broader cyber push? February 5 was the date OpenAI formally introduced Trusted Access for Cyber as a trust-based framework for vetted defenders and committed $10 million in API credits for cyber defense work. (koreaherald.com) On May 7, the company said GPT-5.5 and GPT-5.5-Cyber were being scaled through that system, with GPT-5.5-Cyber reserved for more specialized defensive use cases. OpenAI has framed the program as a way to speed defensive adoption before more capable cyber models become widely available. Kwon said in Seoul that the company did not want advanced cyber capabilities kept “in the hands of a small number of organizations,” and wanted trusted defenders to get the tools faster than bad actors. ### What safeguards come with the access? OpenAI said trusted users still face restrictions on harmful requests even when they receive broader cyber access. (openai.com) The company also said individuals using its most cyber-capable and permissive models through Trusted Access for Cyber will be required to enable Advanced Account Security beginning June 1, 2026. Jason Kwon also said in Seoul that OpenAI offers data residency options so data can be stored in Korea, and for some customers offers a mode in which data is not stored at all, according to Kyunghyang Shinmun. (koreaherald.com) ### What comes next for the Korea rollout? Private-sector expansion is the next step. The Korea Times said OpenAI plans to widen the TAC program to Korean companies in key industries, though company names were not disclosed on Wednesday. (openai.com) June 1, 2026 is the next dated milestone disclosed by OpenAI. That is when Advanced Account Security becomes mandatory for individual trusted users accessing the company’s most cyber-capable models, according to OpenAI’s May 7 policy update. (khan.co.kr) (openai.com) (koreatimes.co.kr)

Key numbers

• OpenAI said on May 27 it expanded its cybersecurity program to South Korea, giving government agencies, public institutions and companies access to defensive AI tools.
• Jason Kwon said Korea will join Governmental Trusted Access for Cyber, with GPT-5.5-Cyber access overseen locally by the Korea Internet & Security Agency.
• From June 1, 2026, OpenAI says trusted users of its most cyber-capable models must enable Advanced Account Security.
• The rollout includes Governmental Trusted Access for Cyber, or GTAC, and access to GPT-5.5-Cyber, a model the company says is tuned for specialized defensive security work.

What happens next

• The Korea Times reported that Korean institutions will gain access to GPT-5.5-Cyber, while OpenAI said the model is being rolled out in limited preview for defenders securing critical infrastructure and other specialized workflows.
• Who in South Korea will run the program?
• (koreatimes.co.kr) The Korea Internet & Security Agency, or KISA, will oversee the program in practice in South Korea, according to the Korea Times.

Sources

• koreaherald.com
• koreatimes.co.kr
• openai.com
• openai.com
• khan.co.kr

Quick answers