Supporting case
India’s cyber agency publishes an AI security audit framework
What happened
On November 4, India’s Computer Emergency Response Team, CERT-In, published guidelines that tell AI operators and auditors how to review systems over their lifetime. The guidance groups systems by risk and sets out evidence an auditor can ask for, including records of model changes, human review and incident handling. For a higher-risk system, an operator may need a formal governance committee and a fuller impact assessment under the audit framework. The document also points to existing incident-reporting directions: the six-hour deadline concerns incidents that those directions already make reportable. A control marked mandatory in the framework is not, by that label alone, a generally enforceable legal duty for every Indian AI user. CERT-In guidelines, including the applicability rules ↗
Common audit evidence can make an operator’s claims easier for purchasers and reviewers to challenge. The distinction is whether a regulator, buyer or contract makes that evidence obligatory and available to someone who can act on it. This guidance and the telecom regulator’s separate October 30 consultation do not establish that India issued the broader AI consultation tracked as D1010-09.
Dates and implementation stage
Action: 2026-11-04; publication: 2026-11-04
