South Korea Joins Anthropic's Project Glasswing to Access Mythos Cybersecurity AI Model

South Korea Gains Access to Advanced Cybersecurity AI

South Korea's Ministry of Science confirmed that the Korea Internet & Security Agency (KISA) has secured access to Anthropic's Mythos AI model through participation in the company's Project Glasswing initiative. The announcement marks a significant step in the country's efforts to strengthen its cyber defenses using frontier AI models designed specifically for identifying and fixing software vulnerabilities 1

1

Source: Korea Times

Anthropics is expanding Project Glasswing to approximately 150 organizations across more than 15 countries, a substantial increase from the initial cohort of roughly 50 institutions. The Ministry of Science and ICT stated it had been working continuously with Anthropic to confirm KISA's participation in this collaborative cybersecurity program built around the Mythos AI model, which was unveiled in April as a security-focused tool designed to identify vulnerabilities at an expert level 2

2

Major Korean Tech Companies Expected to Join

Industry sources indicate that Samsung Electronics, SK Hynix, and SK Telecom are expected to be included as new partners in Anthropic's Project Glasswing, though their participation has not been officially confirmed. Samsung Electronics declined to comment, while SK did not immediately respond to requests for comment . Both Samsung Electronics and SK Hynix previously participated in Anthropic's Series H funding round as strategic infrastructure partners, with their combined investment reportedly in the trillion-won range and Samsung's commitment alone reaching several trillion won 2

2

KISA is expected to be granted access to the Mythos Preview model and related tools such as Claude Security. The state-run agency for cyber incident response and digital security will use Claude Mythos alongside other participating organizations for vulnerability detection, patching, and pre-release security testing 2

2

Addressing Escalating AI-Driven Cyber Threats

Anthropics has maintained strict access controls for the Mythos model due to concerns about potential misuse. The company framed the rapid expansion as a response to escalating AI-driven cyber threats and a shrinking window for controlled deployment. Early participants in Project Glasswing have already used the cybersecurity AI to uncover more than 10,000 high-severity or critical vulnerabilities within weeks 2

2

"Within 6 to 12 months, we expect that many other AI companies will have Mythos-class models, and they could release them without safeguards that prevent misuse," Anthropic stated. "In that world, cyberattacks could occur much more often, and in much more unpredictable forms. It's imperative that cyber defenders adapt to maintain pace" 2

2

Source: Reuters

Related Stories

Anthropic expands Claude Mythos access to 200 partners across critical infrastructure worldwide

Anthropic opens Seoul office and forges partnerships across South Korea's AI ecosystem

OpenAI launches GPT-5.4-Cyber model days after Anthropic's Mythos sparks global security concerns

Security researchers use Anthropic Mythos to expose macOS vulnerabilities on Apple M5 chips

Strengthening National Cyber Defenses

The Ministry of Science emphasized that South Korea would continue efforts to improve its AI-based information security capabilities, including by using various frontier AI models and strengthening domestic technologies. This move positions the country to better defend against sophisticated cyber threats as AI capabilities become more widely available . The participation of KISA and potentially major Korean tech companies signals a comprehensive national approach to leveraging advanced AI for protecting critical infrastructure and digital assets in an era where the threat landscape is rapidly evolving.