AI · Security · 2026-09-11 12:19:00
The Cloud Security Alliance AI Safety Initiative reports that LLMjacking—the hijacking of large language models—has reached black market scale. Attackers steal cloud credentials to gain unauthorized access to LLMs and resell inference capacity for profit. Sysdig's threat research team found stolen credentials circulating in underground markets, with attackers using them to call model APIs while victims bear the costs. The phenomenon spans credential security, cloud security, and financial risk. The report urges enterprises to strengthen credential management, enforce least privilege, and monitor anomalous API calls. The industrialization of LLMjacking shows AI infrastructure is becoming a new target for organized crime.
Source: labs.cloudsecurityalliance.org
Auto-generated summary for reference only. RSS