Korean academic society holds workshop on trustworthy AI and agentic AI security.

Korean academic society holds workshop on trustworthy AI and agentic AI security.
The Korean Institute of Information and Communication Engineering (KIICE) and the AX Trust Research Workshop held the 2026 First AX Trust Research Workshop in Jeju, gathering industry, academic, and government experts to discuss security, industrial, and legal responses to the spread of agentic AI. Topics included Anthropic's Claude Mythos Preview, the Project Glasswing security alliance, sovereign AI strategies, and refinements to the AI Basic Act's overly broad safety definition.
This workshop signals growing institutional attention to the trust and safety layer of the AI stack, a recurring pattern where academic and policy bodies play catch-up to fast-moving foundation-model deployment. The focus on agentic AI security and sovereign AI aligns with cross-substrate forces around safety governance (cross.§G) and geopolitically driven national AI autonomy (cross.§E).
As agentic AI moves beyond tool use into system integration and decision support, the need for access controls, audit trails, and clear liability frameworks becomes acute. The workshop's call for more precise legal definitions reflects an open debate between flexibility and clarity in AI regulation — one that will shape compliance costs for enterprises deploying agentic workflows.