Policies · eGov
India's national cybersecurity agency has released a 107-page guideline requiring organisations that develop or deploy AI systems in India to undergo regular security audits tied to a four-level risk classification.
✦ AI SUMMARY 30-second read
- AI systems are classified as low, moderate, high or critical risk; audit intensity scales accordingly, from self-assessment to continuous monitoring.
- Foreign companies offering AI services or tools to Indian users are covered, and no exemptions apply to smaller firms or startups for high-risk and critical systems.
- Organisations using generative AI must adopt an anti-deepfake policy and label AI-generated media with content provenance markers such as C2PA or an equivalent.
Why it matters: The guidelines extend India's existing cybersecurity incident-reporting rules to AI systems and make audit compliance mandatory across sectors.
This summary was generated by AI. Editors may review and update it.
Document details
| Document type | Guideline |
|---|---|
| Issued by | CERT-In / DoT / TRAI |
| Region | India |
| Published | 5 November 2026 |
The Indian Computer Emergency Response Team (CERT-In) released its AI Security Auditing Guidelines on November 4, requiring any organisation that develops or deploys AI systems in India to carry out regular security assessments. The 107-page document was prepared by a working group at the agency.
The guidelines apply not only to India-based organisations but also to foreign companies that offer AI services or tools to Indian users, or that handle Indian data. CERT-In made clear that compliance standards for high-risk and critical systems are uniform, with no relaxed treatment for smaller firms or startups.
What the document says
The guidelines place every AI system into one of four risk tiers — low, moderate, high and critical — based on factors such as autonomous decision-making capacity, data sensitivity, the number of people affected, and how reversible any harm would be.
Low-risk systems may be assessed through self-assessment or a basic internal review. Moderate-risk systems require an annual independent audit by an auditing organisation empanelled with CERT-In. High-risk and critical systems also require independent audits by empanelled auditors, and critical systems must be monitored continuously.
Where auditors identify two or more critical problems, or issue an unsatisfactory opinion, the organisation must commission a fresh audit within three months. Critical systems must additionally undergo red-team attack exercises every year.
Organisations running AI systems in production must maintain an 'AI bill of materials' — a record of the models, datasets, prompts, tools and software each system relies on, scaled to its complexity and risk level. CERT-In had earlier issued technical guidance on such records; the new guidelines make it a formal audit item.
Generative AI and agentic AI provisions
Organisations that use generative AI must have a policy prohibiting the creation of deepfakes of real people without their consent. They must also mark AI-generated media with content provenance labels such as C2PA or an equivalent standard.
AI systems that act autonomously — described in the document as agentic AI — face additional checks for risks including prompt injection and tool misuse.
Extension of existing rules
The guidelines bring AI-related incidents within the scope of CERT-In's existing reporting requirements, under which reportable incidents must be flagged within six hours and logs retained for 180 days. Compliance with these obligations becomes part of the AI security audit.
The document names the Reserve Bank of India, Securities and Exchange Board of India, IRDAI and the Telecom Regulatory Authority of India as intended users of the guidelines, but sets no sector-specific rules for any of them.
What is not addressed
The document does not specify the legal provision under which the guidelines were issued. It also does not state an effective date or describe the consequences for organisations that do not comply.