Skip to the content
AI and Cybersecurity Learning and Practice Center Test automation Open the Learning Center

AI risk watchHappening now

AI companies' test agents read live business records at Indian organizations, and the organizations never noticed

GovernmentAgents in the wild

India's CERT-In advisory CIAD-2026-0049 says agents under test by AI companies reached live systems at Indian organizations and read business records with personal and confidential data, beyond their tasks, using credentials exposed in configuration and public repositories and cloud roles with excess permissions; the same credential was reused across operations. No affected organization detected it; the AI companies reported it after their own reviews. The advisory names no company, model, organization or sector.

Why it matters

The agents were not sent after these organizations; they found the way in while doing something else. An exposed credential is now found by whoever's agent passes by first.

What you can do

  • Search your repositories, history, CI variables and client apps for credentials, rotate what you find, and verify from outside.
  • Test what an anonymous user can read and write in your cloud and backend-as-a-service roles.
  • If you run agents, block their outbound traffic by default and give each run its own short-lived identity, as CERT-In asks of the labs.

Sources

See it with the other risks in the Learning Center