CVE-2025-67729Deserialization of Untrusted Data in Lmdeploy

Severity
8.8HIGHNVD
EPSS
0.1%
top 78.79%
CISA KEV
Not in KEV
Exploit
No known exploits
Affected products
Timeline
PublishedDec 26

Description

LMDeploy is a toolkit for compressing, deploying, and serving LLMs. Prior to version 0.11.1, an insecure deserialization vulnerability exists in lmdeploy where torch.load() is called without the weights_only=True parameter when loading model checkpoint files. This allows an attacker to execute arbitrary code on the victim's machine when they load a malicious .bin or .pt model file. This issue has been patched in version 0.11.1.

CVSS vector

CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:HExploitability: 2.8 | Impact: 5.9

Affected Packages2 packages

NVDinternlm/lmdeploy< 0.11.1
PyPIinternlm/lmdeploy< 0.11.1

Patches

🔴Vulnerability Details

2
OSV
lmdeploy vulnerable to Arbitrary Code Execution via Insecure Deserialization in torch.load()2025-12-26
GHSA
lmdeploy vulnerable to Arbitrary Code Execution via Insecure Deserialization in torch.load()2025-12-26

🕵️Threat Intelligence

1
Wiz
CVE-2025-67729 Impact, Exploitability, and Mitigation Steps | Wiz