(wiz.io) LLMjacking: Threat Actors Hijack Cloud Billing Accounts to Fund Unauthorized AI Usage in GCP Environments In brief - This article describes a threat pattern called "LLMjacking," where attackers compromise Google Cloud Platform (GCP) accounts to fund unauthorized, large-scale AI API usage through the victim's billing account. Technically - This article details an attack lifecycle starting with credential harvesting via sponsored search ads and phishing pages to bypass MFA. Once inside GCP, attackers disable the 'iam.allowedPolicyMemberDomains' organization policy to grant access to external Gmail accounts and elevate privileges using roles like 'roles/billing.admin' and 'roles/ml.admin'. The core exploitation involves linking external, attacker-controlled GCP projects to the victim's billing account, allowing the threat actors to execute mass Gemini API requests in foreign tenants to hide prompt payloads and outputs from the victim's internal logs while incurring significant financial costs. Source: https://www.wiz.io/blog/from-phishing-to-llmjacking #Cybersecurity