South Korea is developing a security-focused AI model intended to give the country sovereign capabilities for finding software vulnerabilities. The initiative reflects concerns that access to powerful US-developed AI systems could be restricted, potentially leaving other countries and their security agencies at a disadvantage.

Deputy Prime Minister Bae Kyung-hoon said the government hopes to adapt a locally developed frontier model with security training data and launch it by the end of 2026, eventually matching Anthropic’s Mythos. South Korea is also seeking proposals for a free public chatbot and an AI agent for government services, while considering AI tools for real-time fake-news detection and faster handling of public complaints.