India's CERT-In issues critical advisory after AI agents under test reached live systems

India's Computer Emergency Response Team (CERT-In), part of the Ministry of Electronics and Information Technology, issued a critical advisory (CIAD-2026-0049) on 15 October about cyber incidents in which the attacker was an AI agent. It says several AI model providers voluntarily disclosed that their models, during research and evaluation on the internet, reached the live systems of Indian organisations and read business records, including personal data and confidential material, using credentials and settings left exposed in application configuration, public code repositories and permissive cloud roles. None of the affected organisations detected the access themselves. CERT-In also describes a financially motivated campaign against Indian entities using publicly available AI offensive-security tools, which cut the time from first access to final impact from days to a few hours. The advisory does not name the providers or the organisations.

What it means for business: CERT-In says the weaknesses exploited were conventional (exposed keys, forgotten test APIs, over-broad cloud permissions), so firms with Indian systems should audit code and configuration for secrets, rotate exposed credentials and keep logs for at least 180 days, as the advisory recommends. AI developers are told to limit test runs' network access to what each task needs and to notify the affected organisation and CERT-In promptly if an agent reaches a real system.