Main Vulnerability Database Vulnerabilities #VU109603

Deserialization of Untrusted Data in PyTorch - CVE-2025-32434

Published: May 21, 2025 / Updated: April 23, 2026

Vulnerability identifier: #VU109603

CSH Severity: High

CVSS v4: 8.6 [CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:A/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N]

CVE-ID: CVE-2025-32434

CWE-ID: CWE-502

Exploitation vector: Remote access

Exploit availability: Public exploit is available

Vulnerability details

The vulnerability allows a remote attacker to execute arbitrary code on the target system.

The vulnerability exists due to insecure input validation when processing serialized data when loading a model using torch.load with weights_only=True. A remote attacker can trick the victim into loading a specially crafted model and execute arbitrary code on the target system.

Successful exploitation of this vulnerability may result in complete compromise of vulnerable system.

Affected software

PyTorch
Python for Scientific Computing
vLLM
Elasticsearch
Maximo Application Suite - Visual Inspection Component
Maximo Application Suite - Monitor Component

How to mitigate CVE-2025-32434

Install updates from vendor's website.

PyTorch - update to 2.6.0
vLLM - update to 0.8.0
Python for Scientific Computing - addressed in versions 3.2.3, 4.2.3
Elasticsearch - addressed in versions 8.19.8, 9.1.8
Maximo Application Suite - Visual Inspection Component - addressed in versions 9.0.22, 9.1.20, 9.2.1
Maximo Application Suite - Monitor Component - addressed in versions 8.10.25, 8.11.23, 9.0.15, 9.1.5

Links to Public Exploits and PoC-codes

Exploit #12626 - CVE-2025-32434-exploit (A script for exploiting a vulnerability in PyTorch with subsequent RCE in library versions < 2.6.0) (April 23, 2026)

External References

https://github.com/pytorch/pytorch/security/advisories/GHSA-53q9-r3pm-6pq6