South Korea’s Korea Internet & Security Agency (KISA) says it is developing security safety guidelines for “agentic” or autonomous AI systems—software that can perform tasks without constant human direction. The agency announced the work in connection with ensuring safer deployment of such systems.
The reports align on the core point that KISA, South Korea’s key internet security body, is preparing new guidance rather than immediately enforcing a specific rule. While the outlets describe the same initiative, they differ in emphasis: one account frames the effort around building “safety guidelines” for AI agents, and another characterizes it as developing “new security guidelines” for autonomous AI. Neither report provides detailed timelines, specific requirements, or enforcement mechanisms in the available excerpts.
Overall, the coverage indicates a regulatory and risk-management response to faster adoption of increasingly autonomous AI tools, with KISA leading the process to define expected security and safety standards.
