CVE-2026-33625
CWE-400
HIGH
CVSS 8.8
NVD
LMDeploy is a toolkit for compressing, deploying, and serving large language models. Versions 012.1 through 0.12.2 contain a code injection vulnerability in `lmdeploy/pytorch/config.py` line 620 that allows an attacker to execute arbitrary Python code by publishing a malicious HuggingFace model with a crafted `quantization_config.quant_dtype` value. When a user loads the model with lmdeploy, the `quant_dtype` is passed to `eval(f'torch.{quant_dtype}')` without any validation. Version 0.12.3 contains a patch.
CVSS v3.1 Metrikleri
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
Attack Vector
Network
AV:N
Attack Complexity
Low
AC:L
Privileges Req.
None
PR:N
User Interaction
Required
UI:R
Scope
Unchanged
S:U
Confidentiality
High
C:H
Integrity
High
I:H
Availability
High
A:H
AI Güvenlik Analizi
🤖 Yapay Zeka Güvenlik Analizi Hazırlanıyor…
CVSS vektörü ve CWE zayıflık yapısı inceleniyor…
Hedef Keşif Sorguları
CVSS Skoru
HIGH
Zayıflık Türleri (CWE)
CWE-400
Referanslar (3)
İlgili CVE'ler
8.7
CVE-2026-130686
vLLM from 0.6.0 before 0.33.0 lacks a request-wide media budget, allowing remote inference clients to exhaust
7.5
CVE-2026-76417
A vulnerability in Object Linking and Embedding, 2nd Generation (OLE2) file parser for ClamAV could allow an u
7.5
CVE-2026-76515
A vulnerability in the Object Linking and Embedding, 2nd Generation (OLE2) file parser for ClamAV could allow
6.9
CVE-2026-130666
ImageMagick before 7.1.2-35 contains a policy bypass vulnerability that allows attackers to evade resource lim
6.9
CVE-2026-130661
ImageMagick before 7.1.2-35 contains a policy bypass vulnerability in the ASE decoder caused by a missing secu
5.3
CVE-2026-95690
Receiving crafted out-of-sequence DTLS handshake fragments may cause OpenSSL to allocate a full message-length