← Back to Feed

Someone Else Is Using Your AI

September 3, 2026 · Fortinet Threat Research · Severity: MEDIUM

FortiCNAPP has analyzed an AWS LLMjacking incident involving a leaked administrator key that enabled unauthorized access to Amazon Bedrock services. Attackers used the compromised credentials to access and exploit large language model services, potentially generating content and consuming AI resources at the victim's expense. The incident highlights the critical importance of securing cloud administrator credentials and monitoring for unusual AI service usage patterns.

FortiCNAPP analyzes an AWS LLMjacking incident involving a leaked administrator key and unauthorized access to Amazon Bedrock.

      

Key Takeaways

  • FortiCNAPP analyzed an LLMjacking incident where a leaked AWS administrator key enabled unauthorized access to Amazon Bedrock AI services.
  • Attackers exploited compromised cloud credentials to consume LLM resources at the victim's expense, potentially generating malicious content using the hijacked AI services.
  • Cloud security teams should implement strict credential management, monitor AI service usage patterns, and deploy CNAPP solutions to detect LLMjacking attacks.
☕ Buy a Coffee