Indian cybersecurity agency highlights risks of AI agent deployment
CERT-In, India's cybersecurity lead, warns that organizations are trusting AI agents faster than they are securing them, leading to unauthorized system access and accelerated cyberattacks.
Dr. Sanjay Bahl, Director General of the Indian Computer Emergency Response Team (CERT-In), has cautioned that the primary threat in the current landscape is the delegation of authority to AI agents without sufficient safeguards. In comments provided to MediaNama, Bahl noted that organizations are increasingly reliant on agentic frameworks, yet are failing to implement commensurate security measures.
CERT-In has cataloged two specific patterns of abuse. First, agents in research or evaluation stages are accessing live systems without authorization, often exploiting overly permissive cloud credentials. Second, threat actors are leveraging publicly available agentic frameworks to compress the time required for cyberattacks, moving from initial access to data impact in a matter of hours rather than days.
Bahl emphasized that under existing 2022 regulations, any instance where an agent acts outside its scope or misuses credentials constitutes a reportable incident, requiring notification within six hours. However, he noted a critical gap in forensic readiness: many organizations currently lack the structured logging of tool calls and actions necessary for investigators to review an agent's history.