AI Wiki · Mediocre Security
Standard#governance#india

India AI Security Auditing Guidelines

CERT-In's 107-page guidelines with four risk tiers, independent audits for high-risk AI systems and audit areas including prompt injection and excessive agent permissions.

India’s CERT-In published 107-page AI Security Auditing Guidelines on Nov 4 2026, according to The420.in. They sort AI systems into four risk tiers (low, moderate, high, critical) based on the consequence of failure, data sensitivity, how independently the system decides and whether harm can be reversed. High-risk systems are expected to get independent third-party audits; critical-risk systems get more extensive independent assessment and continuous monitoring. The report lists audit areas such as governance, data quality, privacy, model risk, fairness, transparency, cybersecurity, human oversight and third-party accountability (it says nine but lists ten), and threats including prompt injection, data poisoning and excessive AI permissions. AI incidents that meet CERT-In’s reportable criteria must be notified within six hours, and security logs kept for 180 days; both come from existing CERT-In directions rather than being new. It is guidance, not a new AI law. We have not seen the CERT-In document itself.

Timeline

Every dated event in the wiki that involves India AI Security Auditing Guidelines, newest first.

Connections

India AI Security Auditing Guidelines and everything one link away. Drag to move, scroll to zoom, click a node to open it.Open in the full graph →

What links here

Government body CERT-In

Standard ISO/IEC 42001

Concept Prompt injection

Sources

  1. The420.in, “CERT-In Issues AI Security Audit Guidelines, Sets Four Risk Categories”, Nov 4 2026