South Korea's Ministry of Science and ICT said on April 14 it held emergency review meetings related to Anthropic and OpenAI launching projects to use their latest artificial intelligence models for cybersecurity.

Anthropic and OpenAI have launched projects to provide their latest AI models with high-performance security capabilities to partner companies. In particular, concerns were raised that Anthropic's "Mythos" could become a security threat because it can automatically carry out security analysis through agent-based coding and reasoning capabilities.

The government moved to prepare countermeasures. The presidential office's National Security Office ordered an emergency response from relevant ministries across the public, private and military sectors. The science ministry urged companies' chief information security officers to watch for security threats using AI and to conduct urgent security checks.

The science ministry asked companies to share information with the Korea Internet & Security Agency if unusual AI-based attacks occur. It also held an emergency review meeting from 2 p.m. on April 14, chaired by Second Vice Minister Ryu Je-myeong (류제명), with chief information security officers from the three telecom operators and major platform companies. At 5 p.m., the head of the Information Security and Network Policy Office held a review meeting with domestic AI security experts.

Bae Kyung-hoon (배경훈), deputy prime minister and minister of science and ICT, said, "High-performance AI-based cybersecurity services are an opportunity to improve security levels, but if abused they can pose a major risk." He added, "We must be able to strengthen cybersecurity capabilities without exposing domestic companies and infrastructure to threats."