What happened
On 4 November 2026, India's national CERT, CERT-In, published the 'Artificial Intelligence (AI) Security Auditing Guidelines: Framework for Secure, Ethical and Auditable Artificial Intelligence Systems' (available at cert-in.org.in/PDF/AI_Security_Auditing_Guidelines.pdf). The ~107-page framework, produced by Sub-group 03 of the Working Group on 'Indian Standards for Cybersecurity Auditing', establishes a common risk-based, lifecycle-orientated approach to auditing AI systems. It classifies AI systems into four risk tiers (Low/Moderate/High/Critical), defines nine core audit domains (AI governance and accountability; data lifecycle/quality/privacy; model risk management; ethics/bias/fairness; explainability and transparency; security, adversarial resilience and incident reporting; continuous monitoring, drift management and validation; human oversight and decision accountability; third-party and AI supply-chain assurance), adds dedicated requirements for GenAI/RAG/agentic/multimodal/autonomous architectures (prompt injection, excessive agency, data and model poisoning, vector/embedding weaknesses, unsafe tool use, persistent-memory and inter-agent risks), specifies a structured audit methodology, a five-level maturity model, and a Critical/High/Moderate/Low finding classification. It requires AI-related security incidents at government and private entities to be reported to CERT-In within six hours, and aligns with ISO/IEC 42001, ISO/IEC 23894, ISO/IEC 27001, NIST AI RMF, MITRE ATLAS and OWASP guidance.
Why it matters
This is the first national AI-security auditing framework from India's CERT, creating a de-facto compliance benchmark for every Indian government body, critical-infrastructure operator, regulated entity and private organisation that develops, procures or operates AI systems, plus AI systems processing Indian data or serving Indian users. The six-hour incident reporting duty and risk-tiered audit depth impose concrete, auditable obligations, and the explicit agentic/GenAI coverage materially extends existing Indian AI governance guidance (MeitY) and CERT-In's vulnerability blueprint into security-audit territory.
Action needed
For organisations operating in or serving India: obtain the framework, classify AI systems into the four risk tiers, map existing controls to the nine audit domains, and stand up the six-hour AI-incident reporting pathway and evidence trail for agentic/GenAI deployments. Auditors and compliance teams should align AI audit scoping and findings classification to this framework now, since CERT-In-empanelled auditing organisations are expected to apply it.