CERT-In lays down AI security audit norms, defines 4 risk tiers

The Cert-In has issued security audit guidelines for AI systems, classifying them into four risk tiers, low, moderate, high and critical. The risk level will depend on an AI system's autonomous decision-making capacity and reach, determining the stringency of its audit. Organisations using generative AI must also have policies against the creation of deepfakes of real people.