Google identifies LLMjacking attacks generating $1 million in cloud AI charges
Original title: Google: LLMjacking attacks rack up $1 million in cloud AI charges
Researchers observe LLMjacking attacks using stolen API keys leading to fraudulent charges exceeding $100000 per day at some organisations.
Large language model hijacking attacks are generating large fraudulent cloud charges, with some organizations losing more than $100,000 per day. Companies can record costs as high as $1 million due to fraudulent calls using stolen API keys. There are two distinct operational clusters coordinating to compromise and monetize enterprise LLM infrastructure. Attackers sell access to services known as LLM mixers, which provide subscribers with unauthorized access to LLMs. One threat actor leveraged stolen third-party API keys to power the token routing service jet-night.com, passing nearly one million dollars in token costs to the victim organization.
- Attackers use exposed cloud credentials, service account keys, GitHub tokens, active session tokens, and LLM API keys
- Attackers search compromised environment for credentials, logs, databases, and secret managers
- Attackers exploited CVE-2026-40217 in LiteLLM to run arbitrary code on the server
Impact: fraudulent cloud charges from hijacked LLM API keys
- 1.prioritize Identity and Access Management
- 2.secure supply chain
- 3.use centralized secrets platforms
Vulnerabilities mentioned
All| CVE | Vulnerability | CVSS | EPSS | Signals | Published |
|---|---|---|---|---|---|
| CVE-2026-40217 | LiteLLM through 2026-04-08 allows remote attackers to execute arbitrary code via bytecode rewriting at the /guardrails/test_custom_code URI | 8.8 | 16% | PoC |
Indicators of compromise
auto-extracted · verify before use · export allAll| Type | Indicator | Context |
|---|---|---|
| domain | newapi.pro | which sells “unlimited access plans” for $60/month, and the newapi[.]pro service. “Another threat actor that executed more than 50 |
From the original article
3 min read at bleepingcomputer.com
Large language model hijacking attacks, also known as LLMjacking, are generating large fraudulent cloud charges, with some organizations losing more than $100,000 per day. [...]
By Bill Toulas · bleepingcomputer.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.