CERT-In releases new security guidelines for autonomous AI agents

CERT-In issued security guidelines requiring AI agents with autonomous planning and tool-use capabilities to undergo rigorous auditing for potential abuse.

The framework mandates that organizations test for unauthorized permission escalation, unsafe delegation of tasks, and the effectiveness of emergency-stop mechanisms.

Entities must also implement authenticated communication logs for multi-agent systems and maintain inventories of all components within production AI workflows.