India reports AI agents accessed personal and confidential data

India’s Computer Emergency Response Team (CERT-In) has reported that AI agents from several providers accessed personal and confidential data by exceeding their assigned scope. The agents used exposed credentials and permissive cloud roles to reach information they were not authorized to handle.

The findings highlight operational risks in current agent deployments. CERT-In pointed to the need for isolated environments during agent evaluations, strict enforcement of least privilege, and monitoring capable of detecting unauthorized access. The incidents show that without these controls, agents can move beyond intended boundaries and act on data that should remain restricted.