Microsoft has detailed a destructive Azure campaign it attributes to a threat actor it tracks as Storm-3168, the group behind JADEPUFFER, described earlier this year as the first documented agentic ransomware operation. The new research shows the crew moving past data theft into outright resource destruction.
Two compromised service principals in the same tenant did the heavy lifting. One spent about 15 and a half hours enumerating virtual machines, subscriptions, and resource groups, logging more than 300 successful read operations. The second handled further discovery, credential collection, and the destructive phase.
From reconnaissance to wiping the recovery path
The attacker targeted Azure Storage accounts, SQL databases, Key Vaults, Function Apps, virtual machines, and App Services, and went after recovery protection locks, a sign of intent to make restoration hard. Microsoft says the activity began in early June 2026 and points to a broader shift toward AI-orchestrated attacks, where a small crew coordinates complex post-compromise work across a cloud estate at speed.
Defenders can cut exposure by treating workload identities as first-class accounts: enforce least privilege, rotate secrets, watch for unused principals, and protect backup and recovery resources. Exposed credentials stay usable until they are revoked or rotated, so deleting the file that leaked them fixes nothing on its own.
