India's CERT-In Issues AI Agent Security Advisory in 2026

India's Computer Emergency Response Team (CERT-In) has issued a formal security advisory addressing the emerging risks posed by autonomous AI agents, marking the first time the nation's primary cybersecurity body has specifically targeted this technology category. The guidance, officially released on October 24, 2026, comes as Indian enterprises rapidly adopt agent-based AI systems for customer service, business process automation, and decision support functions.

Core Security Concerns Identified

The CERT-In advisory outlines three primary risk vectors associated with AI agent deployment. First, the agency warns that autonomous agents with broad system access can inadvertently exfiltrate sensitive data through their decision-making processes, particularly when integrated with third-party APIs or cloud services. Second, the opacity of agent reasoning chains makes it difficult for security teams to audit decisions or detect when an agent has been compromised. Third, AI agents significantly expand an organization's attack surface by creating new entry points for adversaries who can manipulate agent behavior through prompt injection or goal hijacking techniques.

According to the advisory, agents that operate with elevated privileges or access to critical business systems pose the highest risk. CERT-In specifically flags scenarios where agents can execute code, modify databases, or interact with external systems without real-time human oversight.

Mandated Security Controls

Organizations deploying AI agents in India must now implement several baseline controls under CERT-In guidance:

• Role-based access limitations that restrict agent permissions to minimum necessary functions
• Comprehensive logging of all agent actions and decision points for forensic analysis
• Input validation layers to prevent prompt injection and adversarial manipulation
• Human-in-the-loop checkpoints for high-stakes decisions involving financial transactions or personal data
• Regular security audits of agent behavior patterns and API interactions

The guidance stops short of mandating specific technical implementations, instead providing a risk-based framework that organizations must adapt to their deployment contexts.

Industry Response and Implementation Timeline

Major Indian technology firms and financial institutions have acknowledged the advisory, with several announcing accelerated timelines for agent security reviews. The advisory does not carry immediate regulatory penalties, but CERT-In has indicated that compliance expectations will be integrated into future cybersecurity audits for critical infrastructure sectors. The agency is working with the Ministry of Electronics and Information Technology to potentially formalize these guidelines into binding regulations by mid-2027.

Global Context and Precedent

India's move follows similar actions by cybersecurity agencies in the United States, European Union, and Singapore, all of which have issued guidance on AI agent risks in 2026. However, CERT-In's advisory is notable for its focus on implementation in resource-constrained environments, recognizing that many Indian organizations lack dedicated AI security teams. The guidance includes specific recommendations for small and medium enterprises, including the use of commercial agent security platforms and third-party monitoring services.

What This Means

CERT-In's advisory signals that Indian regulators view autonomous AI agents as a distinct security category requiring specialized controls beyond general AI governance frameworks. Organizations operating in India should prioritize agent security architecture reviews and establish clear accountability chains for agent-driven decisions. As AI agents become more capable and widely deployed, this guidance likely represents the first wave of more comprehensive regulatory requirements that will emerge across 2027 and beyond. Companies that proactively address these concerns will be better positioned for compliance and customer trust in an increasingly agent-driven business environment.