Enterprises pay $100,000 a day for AI they never ordered
Summary
Enterprises deploying LLMs are vulnerable to "LLMjacking" campaigns where cybercriminals steal cloud and model access credentials. These stolen credentials are then resold through underground subscription services, leading to potentially massive, unexpected cloud charges for the victimized organizations.
IFF Assessment
This article highlights a new threat where attackers exploit enterprise AI infrastructure for financial gain, negatively impacting defenders by creating significant financial and security risks.
Severity
Defender Context
Defenders need to be aware of LLMjacking campaigns, which involve the theft and reselling of enterprise AI model access. This tactic can lead to substantial financial losses and compromise sensitive data if not properly secured. Organizations should implement robust credential management, monitor cloud spending for unusual spikes, and enforce strict access controls for AI models and cloud resources.