Observed incident

Phished cloud accounts fund unauthorized Gemini usage

Reported

Published here

What happened

Wiz investigated multiple GCP intrusions in which phished privileged accounts were used to attach attacker-controlled projects to victims' billing accounts, generating tens to hundreds of thousands of dollars in unauthorized Gemini API charges within 24–48 hours.

Parties and reporting

AI tools and systems: Google Cloud billing and IAM, Gemini API, Search-ad phishing kit

Outcome

Google Ads removed the reported malicious ad campaign. Victim recovery, reimbursement and campaign-wide containment were not disclosed; prompts and model outputs remained outside victims' visibility.

Operating lesson

Use phishing-resistant MFA, restrict billing and organization-policy privileges, alert on external project links and monitor anomalous AI spending.

Evidence and limits

Observed resource theft, not proof AI conducted the phishing or that Gemini was compromised. Actor identities, victim geography and exact incident dates are undisclosed. The purpose of the hidden AI workloads is unknown.

Sources