What is the CVSS score of CVE-2025-33244?

CVE-2025-33244 has a CVSS 3.1 base score of 9.0 (Critical). CVSS vector: CVSS:3.1/AV:A/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H. EPSS exploitation probability: 0.0%.

Which versions of Pytorch are affected by CVE-2025-33244?

A critical vulnerability in NVIDIA APEX affects PyTorch environments, enabling attackers on adjacent networks to execute arbitrary code, escalate privileges, and access sensitive data without a patch…

How to fix or mitigate CVE-2025-33244?

Within 24 hours: Inventory all systems running NVIDIA APEX with PyTorch <2.6 and isolate critical instances from network access where feasible. Within 7 days: Upgrade PyTorch to version 2.6 or later on all applicable systems; apply network segmentation to restrict adjacent network access to APEX-dependent services; implement monitoring for suspicious deserialization activity. Within 30 days: Complete full remediation across all environments, conduct security assessment of affected systems for signs of exploitation, and establish automated inventory tracking for NVIDIA APEX and PyTorch versions.

Pytorch CVE-2025-33244

EUVD-2025-208972 CRITICAL

Deserialization of Untrusted Data (CWE-502)

2026-03-24 nvidia

Information Disclosure RCE Deserialization Denial Of Service Nvidia Pytorch AI / ML

9.0

CVSS 3.1 · NVD

Severity by source

NVD PRIMARY

9.0 CRITICAL

AV:A/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H

Primary rating from NVD · only source for this CVE.

CVSS VectorNVD

CVSS:3.1/AV:A/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H

Attack Vector

Adjacent

Attack Complexity

Low

Privileges Required

Low

User Interaction

None

Scope

Changed

Confidentiality

High

Integrity

High

Availability

High

Lifecycle Timeline

EUVD ID Assigned

Mar 24, 2026 - 20:31 euvd

EUVD-2025-208972

Analysis Generated

Mar 24, 2026 - 20:31 vuln.today

CVE Published

Mar 24, 2026 - 20:25 nvd

CRITICAL 9.0

DescriptionCVE.org

NVIDIA APEX for Linux contains a vulnerability where an unauthorized attacker could cause a deserialization of untrusted data. This vulnerability affects environments that use PyTorch versions earlier than 2.6. A successful exploit of this vulnerability might lead to code execution, denial of service, escalation of privileges, data tampering, and information disclosure.

AnalysisAI

NVIDIA APEX for Linux contains a deserialization of untrusted data vulnerability that affects environments using PyTorch versions earlier than 2.6. An attacker with low privileges on an adjacent network can exploit this flaw to achieve code execution, denial of service, privilege escalation, data tampering, and information disclosure with scope change (CVSS 9.0 Critical). No KEV listing or public POC availability has been reported at this time.

Technical ContextAI

NVIDIA APEX (A PyTorch Extension) is a tool for mixed precision and distributed training in PyTorch environments, commonly used to accelerate deep learning workloads. The vulnerability stems from CWE-502 (Deserialization of Untrusted Data), where the application fails to properly validate serialized data before processing it. This affects the cpe:2.3:a:nvidia:apex product across all versions when used with PyTorch versions prior to 2.6. Deserialization vulnerabilities occur when applications deserialize objects from untrusted sources without sufficient validation, allowing attackers to inject malicious serialized objects that execute arbitrary code during the deserialization process.

RemediationAI

The primary remediation is to upgrade PyTorch to version 2.6 or later, which mitigates the deserialization vulnerability in NVIDIA APEX environments, as detailed in the NVIDIA security advisory at https://nvidia.custhelp.com/app/answers/detail/a_id/5782. Organizations unable to immediately upgrade should implement network segmentation to restrict adjacent network access to APEX deployments, enforce strict authentication and authorization controls to limit low-privilege account access, and monitor for suspicious deserialization activities or unexpected process execution from APEX components. Consider implementing application-level input validation and restricting the sources from which serialized data is accepted until the PyTorch upgrade can be completed.