India’s CERT-In Releases AI Security Auditing Guidelines
India’s CERT-In has issued security auditing guidelines for AI systems. The guidance focuses on risks including model poisoning, prompt injection and threats involving agentic AI.
For organizations deploying AI, the guidelines provide a framework for evaluating systems against attacks that can manipulate models or influence their behavior. As AI agents gain greater autonomy and access to tools or workflows, auditing their security controls is becoming an important part of system assurance.