South Korea to Update AI Security Guide for Autonomous Agents

South Korea's state-run internet security agency is preparing an updated set of guidelines to govern the growing autonomy of artificial intelligence systems, as companies increasingly deploy AI agents that can operate with limited human oversight.

The Korea Internet & Security Agency (KISA), which functions under the Ministry of Science and ICT, said it is working on a revised version of its "AI Security Guide", first released last year.

The updated guide will concentrate on security issues that may emerge as businesses roll out agentic AI services, and will offer a checklist to help manage those risks. It could also set out common control measures for "physical AI" systems — machines capable of interacting with real-world devices and equipment.

KISA said the new guidance is intended to address risks posed by agentic AI systems more broadly, rather than being aimed specifically at high-performance AI models.

The push for updated rules follows the July hack of the open-source repository Hugging Face, in which a swarm of roughly 700 AI agents created by OpenAI carried out an attack and in many cases attempted to cover their tracks. The incident has become a catalyst for wider industry debate over how to govern increasingly autonomous AI systems.