Story perspectives

AI Agents Breached Indian Systems in Hours, CERT-In Warns

By Drooid · Oct 16, 2026, 7:57 AM UTC · How we work

Summary

Story summary

Autonomous AI agents accessed live Indian organisational systems, reading business records and personal data without authorization.
The agents exploited exposed passwords, digital keys, credentials in public code repositories, permissive cloud roles, and dormant APIs lacking MFA.
CERT-In advised firms to audit credentials, retire unused APIs, enforce least-privilege access and multi-factor authentication.
Cybercriminals used open-source AI agents, reducing intrusion-to-impact time from days to a few hours.

History

Historical Context

In May 2017 the WannaCry ransomware used the NSA-derived EternalBlue exploit, leaked by the Shadow Brokers, to encrypt more than 300,000 Windows machines in 150 countries; the NHS alone lost 80,000 appointments before researcher Marcus Hutchins registered a kill-switch. That rapid, credential-free spread now mirrors today’s AI agents that harvest public keys, prompting zero-trust and MFA adoption.

1 of 2

Story summary

Autonomous AI agents accessed live Indian organisational systems, reading business records and personal data without authorization.
The agents exploited exposed passwords, digital keys, credentials in public code repositories, permissive cloud roles, and dormant APIs lacking MFA.
CERT-In advised firms to audit credentials, retire unused APIs, enforce least-privilege access and multi-factor authentication.
Cybercriminals used open-source AI agents, reducing intrusion-to-impact time from days to a few hours.

1 / 2