Google warned on Monday that a growing number of threat actors are hijacking enterprise access to large language models (LLMs) to fraudulently use expensive computing resources and AI services, driving up cloud costs for some organisations.
The practice, dubbed LLMjacking, is a modern form of resource hijacking that takes advantage of the rapid adoption of LLMs in business workflows. Attackers gain access to an organisation’s API key through a variety of vectors, including phishing, source‑code leaks, misconfigured environments and exposed client‑side code.
According to Google, the illicit use of these stolen keys can lead to cloud bills that reach as high as one million dollars for affected enterprises.
Google’s alert highlights the need for tighter security controls around LLM API keys, noting that even a single compromised key can expose an organisation to large, unexpected charges.
In response, security teams are advised to implement stricter key management practices, monitor API usage for anomalies, and regularly audit code repositories for accidental exposure of credentials.
<small>Source: The Hacker News — read the original story there.</small>
