Seoul plans domestic security-focused AI models and global cooperation as Claude Mythos Preview raises concerns over AI-driven cyberattacks
South Korean Minister of Science and ICT and Deputy Prime Minister for Science Affairs Bae Kyung-hoon (C) speaks during a meeting of science and technology ministers at the government complex in Seoul, South Korea, 30 April 2026. Photo by YONHAP / EPA
South Korea is preparing new measures to counter AI-powered cyberattacks after Anthropic’s Claude Mythos Preview demonstrated advanced ability to find and exploit software vulnerabilities.
The Ministry of Science and ICT plans to announce countermeasures as early as late May, officials said Tuesday.
The move follows what officials described as the “Mythos shock,” after Anthropic said its unreleased frontier model could outperform all but the most skilled humans in discovering software vulnerabilities. Anthropic has limited the model’s use to defensive cybersecurity work through Project Glasswing, a partnership that includes Amazon Web Services, Apple, Google, Microsoft, NVIDIA and other major technology firms.
Science and ICT Minister Bae Kyung-hoon said AI is reshaping cybersecurity and that South Korea must build a security system capable of defending against AI with AI.
The ministry and the Korea Internet & Security Agency recently used publicly available AI models to simulate an attack on corporate services and found seven vulnerabilities in about 10 minutes, officials said.
The planned response is expected to include stronger AI-based cyber defense capabilities and development of domestic AI models specialized for Korean security conditions. Officials said the goal is to reduce reliance on foreign models while building systems suited to local networks and infrastructure.
The ministry is also expanding cooperation with global AI companies. On Monday, Second Vice Minister Ryu Je-myung met Michael Sellitto, Anthropic’s head of global affairs, and agreed to continue cooperation on cybersecurity uses of AI.
South Korea also expressed interest in participating in Project Glasswing, which Anthropic launched to help secure critical software against emerging AI-era threats.
— Reported by Asia Today; translated by UPI
© Asia Today. Unauthorized reproduction or redistribution prohibited.
Original Korean report: https://www.asiatoday.co.kr/kn/view.php?key=20260512010003082
