CVE-2026-41523
EUVD-2026-3840622.06.2026, 23:16
vLLM is an inference and serving engine for large language models (LLMs). Prior to 0.22.0, an assert-based security check in vLLM's activation function loading allows any unauthenticated attacker to achieve arbitrary code execution on the server by publishing a malicious HuggingFace model, when vLLM runs in Python optimized mode (python -O or PYTHONOPTIMIZE=1). This vulnerability is fixed in 0.22.0.
Affected Products (NVD)
| Vendor | Product | Version |
|---|---|---|
| vllm | vllm | 𝑥 < 0.22.0 |
𝑥
= Vulnerable software versions
Early Detection
Affected products identified ahead of NVD analysis through intelligence sources.
| Vendor | Product | Version | Source |
|---|---|---|---|
| Red Hat | Red Hat AI Inference Server 3.2 | 1782951012 ≤ 𝑥 < * | ADP |
| Red Hat | Red Hat AI Inference Server 3.2 | 1782951244 ≤ 𝑥 < * | ADP |
Common Weakness Enumeration
- CWE-94 - Improper Control of Generation of Code ('Code Injection')The software constructs all or part of a code segment using externally-influenced input from an upstream component, but it does not neutralize or incorrectly neutralizes special elements that could modify the syntax or behavior of the intended code segment.
- CWE-617 - Reachable AssertionThe product contains an assert() or similar statement that can be triggered by an attacker, which leads to an application exit or other behavior that is more severe than necessary.
References