PT-2026-28373 · Unknown+1 · Hugging Face+1
Wernerina
·
Published
2026-03-26
·
Updated
2026-03-29
·
CVE-2026-27893
CVSS v3.1
8.8
High
| Vector | AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H |
Name of the Vulnerable Software and Affected Versions
vLLM versions 0.10.1 through 0.17.x
Description
vLLM is an inference and serving engine for large language models (LLMs). Starting with version 0.10.1 and continuing through version 0.17.x, two model implementation files hardcode
trust remote code=True when loading sub-components. This bypasses the user’s explicit --trust-remote-code=False security setting, potentially enabling remote code execution via malicious model repositories, even when the user has disabled remote code trust. The vulnerability occurs because the system does not respect the user-defined security opt-out. The affected files override the user's setting without any warning or log entry. A malicious Hugging Face repository targeting either architecture can achieve code execution on the inference server.Recommendations
Versions 0.10.1 through 0.17.x are vulnerable and should be updated to version 0.18.0 or later.
Exploit
Fix
Protection Mechanism Failure
Found an issue in the description? Have something to add? Feel free to write us 👾
Weakness Enumeration
Related Identifiers
Affected Products
Hugging Face
Vllm