CVE-2026-41523
7.5
Vector
CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:H/A:H
Exploitability: 1.6 / Impact: 5.9
Source: security-advisories@github.com (Secondary)
Description
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 (1)
Related CWEs
CWE-617
Reachable Assertion
The 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.
CWE-94
Improper Control of Generation of Code ('Code Injection')
The product 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.
References (8)
Source: security-advisories@github.com
Patch
Source: security-advisories@github.com
ExploitThird Party Advisory
Source: security-advisories@github.com
Third Party Advisory
Source: 0b0ca135-0b70-47e7-9f44-1890c2a1c46c
Source: 0b0ca135-0b70-47e7-9f44-1890c2a1c46c
Source: 0b0ca135-0b70-47e7-9f44-1890c2a1c46c
Source: 0b0ca135-0b70-47e7-9f44-1890c2a1c46c
Source: 0b0ca135-0b70-47e7-9f44-1890c2a1c46c
Timeline
No history available yet.