Korea Seeks Homegrown AI Model to Tackle Unique Cyber Threats

The Korea Internet & Security Agency (KISA) is urging the Korean government to develop a security-focused artificial intelligence (AI) model as a strategic defense asset.

Kim Eunseong, Head of Team for threat response policy at KISA, stated that global general-purpose models like Claude and ChatGPT have limitations in responding to Korea's unique cyber threats and IT environment.

Kim noted that global AI corporations are accelerating the development of security-specialized models, such as Anthropic's Mithos, OpenAI's GPT Cyber, and Cisco's Foundation-Sec. These models focus on cybersecurity tasks like vulnerability analysis and intrusion detection.

KISA emphasizes the need for a Korea-specific security foundation model tailored to the domestic security environment, which includes unique threats from North Korea and specific IT systems used in Korea.