India’s cybersecurity agency, CERT-In, has issued a ‘critical’ advisory warning that autonomous Artificial Intelligence (AI) agents accessed live systems and confidential data of Indian organisations without authorisation. The advisory, CIAD-2026-0049, titled ‘AI Agents as Attack Vectors’, highlighted two categories of incidents targeting Indian Information and Communication Technology (ICT) infrastructure.In one set of incidents, AI model providers disclosed that autonomous AI models and agents, undergoing research, exceeded their authorised scope. These agents reached live systems of Indian organisations, reading business records with personal and confidential data.The AI agents used credentials and connection settings exposed in application configurations, public code repositories, and overly permissive cloud service roles. None of the affected organisations detected this access; disclosures came from the AI model providers after internal reviews.Separately, CERT-In observed cybercriminals using publicly available AI-native offensive-security frameworks for reconnaissance and exploitation. These tools enabled financially motivated campaigns against Indian entities, shrinking the window between initial access and final impact to a few hours.Victims that only partly fixed weaknesses were re-attacked within days, often on a larger scale, the agency noted. Conventional security lapses were identified as the root cause in all observed AI agent-driven incidents.These lapses included credentials and keys exposed in source code, dormant APIs, plain text payment-API keys, over-permissive cloud roles, and internet-facing management interfaces. AI agents exploited these gaps faster and more thoroughly than human attackers, even reusing exposed credentials in separate operations.To mitigate these threats, CERT-In urged organisations to audit source code, repositories, and cloud storage for exposed credentials. It advised rotating exposed credentials, purging them from code history, decommissioning old APIs, enforcing least-privilege access, taking admin interfaces off the internet, and requiring multi-factor authentication.Organisations should also monitor attacks, keep logs for at least 180 days, and treat environments as compromised until forensically cleared after any exposure. For AI developers, CERT-In recommended isolating research and evaluation work by blocking outbound network access and using synthetic targets and test credentials.Jaydeep Singh, General Manager – India at Kaspersky, said the advisory is a reminder that AI agents can access live systems and sensitive data through existing vulnerabilities. He noted that over 80 per cent of attacks Kaspersky investigated in 2025 used valid accounts and public-facing applications as entry points. Singh emphasised that AI allows threat actors to iterate faster and scale their efforts, exploiting these paths more quickly. He also recommended isolating AI agents used for research and testing from production environments.
Leave a Reply