India's CERT-In issues AI security auditing guidelines with four risk tiers and six-hour AI incident reporting

CERT-In · Guidelines

Read the source (opens in a new tab)
What happened
India's Computer Emergency Response Team published AI Security Auditing Guidelines, version 1.0 dated November 4, 2026, for AI systems operated in India, processing Indian data or serving Indian users.
Key points
  • Each system is scored on eight risk factors, including autonomy and reversibility, into Low, Moderate, High or Critical tiers that set audit depth, frequency and independence.
  • High-tier systems get periodic third-party audits by CERT-In empanelled auditors; Critical-tier systems need an "Independent audit by CERT-In empanelled auditing organisation with continuous monitoring."
  • "AI incidents meeting CERT-In reportable criteria shall be reported within 6 hours."
  • "High-risk agentic systems shall support emergency interruption, credential revocation, rollback and prevention of uncontrolled automatic restart."
Evidence
Official guidelines from the national cyber agency, prepared by a CERT-In working group. The 107-page document sets audit requirements and checklists for in-house, purchased and API-based AI systems.
Uncertain
The document names no compliance deadline and no penalties for systems that are not audited.

See it in the latest record