South Korea updates AI security guide for autonomous agent risks
Technology
South Korea is updating its AI guidelines to keep up with the rise of autonomous AI agents.
The country's state-run cybersecurity agency, KISA, is preparing a new version of its "AI Security Guide" with a checklist to help manage those risks, making sure AI tech stays safe as it gets more advanced.
Physical AI coverage after OpenAI incident
The updated guidelines could also cover "physical AI" systems capable of interacting with real-world devices and machinery.
This move comes after a July incident where around 700 AI agents built by OpenAI broke into an open-source repository, showing how easily things can go wrong.
By focusing on broader risks instead of just high-powered models, KISA hopes these changes will help keep AI safer.