CERT-In’s AI Audit Guidelines Set Risk Tiers, Independent Audits and Reporting Requirements

· Original article ↗

Summary

Cyble explains CERT-In’s AI security audit guidelines, including eight-factor risk scoring, tier-based audits, systems in scope, and the six-hour incident-reporting and 180-day log-retention requirements.

Key points

  • The guidelines cover AI systems developed, procured, deployed, or operated in India, as well as systems processing Indian data or serving Indian users, regardless of model origin.
  • Systems are scored from 1 to 4 across eight risk factors; total scores from 8 to 32 determine one of four audit tiers.
  • Tier 2 and higher require an independent audit by a CERT-In-empanelled organization; audit depth and frequency increase at higher tiers.
  • Autonomous actions such as issuing refunds can raise an AI system’s risk score and may trigger escalation to a higher tier.
  • The article says qualifying AI incidents must be reported within six hours, with logs retained for 180 days, under CERT-In’s 2022 Directions.
  • Recommended preparation includes inventorying AI systems, scoring their actual capabilities, reviewing permissions, securing credentials, and engaging an auditor early.

Article Details

Topic
CERT-In’s AI Security Auditing Guidelines, including risk-tier scoring, audit requirements, and incident reporting obligations.

Countries