CVE Tools

Google Warns LLMjacking Attacks Are Costing Some Enterprises Up to $1 Million in Cloud Charges

The Hacker NewsBy The Hacker News

Our summary

Google's Threat Intelligence Group and Mandiant report that attackers are increasingly hijacking stolen credentials and API keys to abuse enterprise access to Large Language Models, a scheme dubbed LLMjacking. Victims across financial services, retail, technology, and transportation sectors in North America, South America, and Asia Pacific have seen fraudulent usage charges exceed $100,000 in a single day, with totals reaching as high as $1 million before access was cut off. Attackers typically gain entry via leaked IAM credentials, Service Account keys, GitHub tokens, or session tokens — one group tracked as UNC6993 even built a fake login portal to steal active session tokens and bypass MFA. Once inside, intruders escalate privileges, harvest secrets, and in some cases exploit a vulnerability, CVE-2026-40217, in the LiteLLM routing software, before reselling stolen model access through underground "LLM Mixer" services like Jet Night. A related report from Symantec/Carbon Black describes a separate grey-market relay that reconfigures legitimate OpenAI Codex and Anthropic Claude Code CLIs to funnel traffic through third-party servers, mainly affecting organizations in China and nearby countries since early 2026.

Read at The Hacker News

The Hacker News publishes this story on its own site; we link to it rather than reprint it.

Worried this affects your company?

Discuss a security assessment of your internet-facing systems. Scope agreed before testing.

Check my exposure