South Korea's National Artificial Intelligence Strategy Committee said it held the first regular meeting of its Security Special Committee on Wednesday afternoon to discuss security threats linked to the recently emerging 'Claude Mythos' and measures to improve rules on installable security software in the financial sector.

The committee said cyber security conditions are changing rapidly as generative artificial intelligence (AI) and AI agents advance and spread in use, increasing the need to review existing security systems overall.

The meeting first carried out an emergency review of developments related to Anthropic's 'Claude Mythos'. The committee explained that a new type of security threat could be possible, such as AI identifying vulnerabilities that people have not found for a long time.

The meeting also discussed improvements to installable security software, which has been criticised for causing inconvenience to users in the financial sector and for potentially being abused as a new attack route. This includes security module management programs and keyboard security programs.

The Security Special Committee, together with relevant ministries including the Ministry of Science and ICT and the Financial Services Commission, shared information on Claude Mythos' performance and threat level, domestic response capabilities, and views from companies and experts, and reviewed response directions.

Members reached a consensus that measures are needed beyond short-term responses such as inspections of national critical infrastructure and strengthening supply-chain security. They said this includes building AI-based real-time defence systems, linking the independent AI foundation model project to strengthening security capabilities, and expanding global security cooperation frameworks.

The committee plans to include relevant agencies such as the National Intelligence Service, the Ministry of Foreign Affairs and the Financial Services Commission, in addition to the Ministry of Science and ICT, at its next meeting to discuss more specific response measures.

Lee Won-tae (이원태), chair of the Security Special Committee at the National AI Strategy Committee, said, "This is now an era in which AI, not people, can take the initiative in new hacking." He added, "If we cannot change security policies in line with the speed of technological evolution, security could become an obstacle to the AI transformation and to becoming a leading AI nation." He said the committee would take the lead in identifying effective policy agendas and work with relevant ministries to help build a national security system suited to the AI era.