CERT-In Urges Testing of AI Agents for Misuse
India's cyber security agency CERT-In has issued guidelines asking organisations to test AI agents for unauthorised tool use and privilege abuse. The checks cover what permissions agents have, how tasks are handed between agents, and whether there is a way to stop them in an emergency. Systems with several agents must keep authenticated, logged communications and records of actions.
Context
Context: AI agents are programs that take actions on their own, like using apps or tools for a user. CERT-In is India's national cyber security agency, which sets security advice for organisations. As companies give agents access to company data, control over what they can do matters more.
Burning Potato score: 72%