South Korean authorities publish zero-trust security guide for agentic AI
South Korea’s Ministry of Science and ICT and Korea Internet & Security Agency announced the publication of a zero-trust security guide for companies deploying and operating agentic AI. The guide calls for verification and security-policy enforcement across agents’ input processing, tool calls, task delegation and external communications, with human approval or preapproved policies depending on risk. It also provides security-maturity assessment criteria and phased implementation procedures.