South Korea expands AI cybersecurity measures after Anthropic warning

According to UPI (reporting Asia Today), South Korea's Ministry of Science and ICT is preparing countermeasures to address AI-powered cyberattacks and plans to announce measures as early as late May, officials said. The announcement follows what officials described as the "Mythos shock," after Anthropic said the unreleased model could outperform all but the most skilled humans at finding software vulnerabilities, per UPI. The ministry and the Korea Internet & Security Agency reportedly used publicly available AI models to simulate attacks on corporate services and found seven vulnerabilities in about 10 minutes, according to officials quoted by UPI.

Per UPI, Anthropic has limited the use of the Claude Mythos Preview to defensive cybersecurity work under Project Glasswing, a collaboration that UPI says includes Amazon Web Services, Apple, Google, Microsoft, NVIDIA and other firms. UPI reports the South Korean government plans to strengthen AI-based cyberdefense capabilities and develop domestic AI models aimed at Korean network and infrastructure conditions, with the stated objective of reducing reliance on foreign models. UPI also reports that Second Vice Minister Ryu Je-myung met Anthropic's Michael Sellitto and agreed to continue cooperation on cybersecurity uses of AI.

Editorial analysis: Governments and national security agencies worldwide have accelerated AI-specific cybersecurity work after frontier models demonstrated rapid vulnerability discovery. Observers following the sector note a pattern where simulated red-team exercises using public models expose high-impact bugs quickly, which in turn pushes states to consider localized defensive tooling, regulatory guardrails, and public-private collaboration.

Editorial analysis: Practitioners and security teams should monitor:

• •whether South Korea publishes technical standards or benchmarks for AI-assisted vulnerability discovery
• •progress on domestic model development tuned to Korean infrastructure
• •any outputs from Project Glasswing or partnerships that provide vetted defensive models or tooling. Attention points include responsible-use controls, disclosure coordination timelines, and how simulated-exploit findings are validated and remediated

All reported facts above are from UPI, which cites reporting by Asia Today and statements attributed to South Korean officials and agencies.