vLLM allows a malicious model RCE by torch.load in hf_model_weights_iterator (vllm-project vllm)

CVE-2025-24357
HIGH 7.5

En bref

Source
NVD
EPSS
EPSS 0.87 % · centile 59 ()
État
PUBLISHED
Publiée
Mise à jour
Attribuée par
GitHub_M
Statut NVD
Analyzed

Description

vLLM is a library for LLM inference and serving. vllm/model_executor/weight_utils.py implements hf_model_weights_iterator to load the model checkpoint, which is downloaded from huggingface. It uses the torch.load function and the weights_only parameter defaults to False. When torch.load loads malicious pickle data, it will execute arbitrary code during unpickling. This vulnerability is fixed in v0.7.0.

Faiblesses (CWE)

  • CWE-502 — Désérialisation de données non fiables

Références

Scores

VersionScoreVecteurÉvalué par
CVSS 3.1HIGH 7.5CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:H/A:HGitHub_M
CVSS 3.1HIGH 7.5CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:H/A:HNVD · security-advisories@github.com (Secondary)
CVSS 3.1HIGH 8.8CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:HNVD · nvd@nist.gov (Primary)

Avis

Références européennes

Avis du CERT-FR

Configurations affectées (CPE NVD)

Produits affectés

ProduitVersions
vllm-project · vllmaffected < 0.7.0

Sources

SourceLicence
CVE List (CVEProject cvelistV5)CVE Program Terms of Use
EPSS (FIRST.org)FIRST — attribution requested
ENISA EUVDENISA — reproduction authorised, source acknowledged
CERT-FR (ANSSI)Licence Ouverte / Etalab 2.0
NVD (NIST) via fkie-cad/nvd-json-data-feedsPublic domain — this product uses data from the NVD API but is not endorsed or certified by the NVD