Korea AI security committee chair proposes A-CSF, "sovereign AI 2.0"
At a DailySecu CISO breakfast seminar in Seoul, the security committee chair of Korea's national AI strategy body called for a Korean AI cybersecurity framework.
The chair of the security special committee at South Korea's National AI Strategy Committee proposed building a Korean AI cybersecurity framework (A-CSF) and moving toward what he called "sovereign AI 2.0" in a lecture on October 29, 2026, according to dailysecu.com. The address was delivered at a CISO breakfast seminar hosted by DailySecu at the Lotte Hotel Seoul in Sogong-dong, Seoul, under the theme of Korea's AI security policy direction and tasks from an AI geopolitics perspective.
The chair said the center of AI competition is shifting from the speed of performance development to risk control and to securing control rules first. Rather than simply owning AI models, he argued, countries need the ability to independently verify, control and operate them — the core of his "sovereign AI 2.0" proposal.
As the key technology, he pointed to a "security harness" that governs the system and data access rights and the execution scope of AI models and agents. He proposed expanding adoption of the harness in the short term and building the A-CSF over the medium and long term.
He named two "security debts" as urgent tasks: identifying assets and managing the attack surface, and poor maintenance and security updates. He also called for institutional measures that reflect AI data center security from the design stage and guarantee a minimum share of AI project budgets for security, and proposed a "Korean version of a 120-day plan" to strengthen AI risk management and resilience in the superintelligence era.
Korea should preemptively establish AI incident classification and assessment standards and a security framework to lead international discussions, the chair said. According to the article, domestic AI security policy has so far focused on generative AI and large language models, with insufficient consideration of the attack capabilities of autonomous AI agents and system resilience, while the government is expanding AI data centers faster than the security policy to protect them.
Quick answers
What is the A-CSF?
The Korean AI cybersecurity framework proposed by the chair of the National AI Strategy Committee's security special committee as a medium- to long-term goal.
What is a "security harness"?
A technology that controls the system and data access rights and execution scope of AI models and agents, which the chair proposed expanding in the short term.
Where was the lecture held?
At the DailySecu CISO breakfast seminar at the Lotte Hotel Seoul in Sogong-dong, Seoul, on October 29, 2026.