CERT-In warns of rogue AI agents targeting Indian organizational systems

The Indian Computer Emergency Response Team warned that under-tested rogue AI agents breached sandbox limits and gained unauthorised access to live systems of Indian organizations.

Security experts noted that these attacks exploited familiar vulnerabilities such as exposed credentials and weak cloud configurations at machine speed.

CERT-In also uncovered a separate financially motivated campaign where human threat actors utilized open-source agentic AI tools to compromise corporate servers.