Stolen cloud identities can produce fraudulent AI charges exceeding $100,000 in one day and cumulative costs of $1 million before access is revoked. Multiple incidents in 2026 affected financial services, retail, technology, and transportation organizations across three global regions. The dominant entry route involved exposed or stolen identity tokens, service account keys, personal access tokens, and active sessions. Attackers enabled AI services, created new accounts and keys, provisioned computing resources, and resold unauthorized access through illicit routing services.
Why It Matters
Any organization using cloud-hosted models or holding credentials that can enable them is exposed. Small validation requests may let attackers test stolen keys without triggering volume alerts. For leaders: AI usage needs spending limits, ownership, and the same identity governance as other expensive cloud services. For defenders: monitor service enablement, key creation, quota changes, unusual requests, and abrupt billing spikes. The broader lesson is that cloud identity compromise can become an immediate financial event even without data encryption or operational downtime.
What Defenders Should Do
Set granular AI quotas and budget alerts while rotating exposed credentials and restricting who can enable model services.