Which versions of Pytorch are affected by CVE-2026-4538?

Organizations using A vulnerability should be aware of moderate risk from CVE-2026-4538, a insecure deserialization vulnerability rated CVSS 5.3.

Is there a public PoC exploit available for CVE-2026-4538?

Yes, a public proof-of-concept exploit is available for CVE-2026-4538. Prioritise patching immediately. EPSS: 0.0%.

How to fix or mitigate CVE-2026-4538?

Within 30 days: Identify affected systems and apply vendor patches as part of regular patch cycle. Monitor vendor channels for patch availability.

Pytorch CVE-2026-4538

Q: What is the CVSS score of CVE-2026-4538?

CVE-2026-4538 has a CVSS 4.0 base score of 1.9 (Low). CVSS vector: CVSS:4.0/AV:L/AC:L/AT:N/PR:L/UI:N/VC:L/VI:L/VA:L/SC:N/SI:N/SA:N/E:P/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X. EPSS exploitation probability: 0.0%.

EUVD-2026-14280 LOW

Deserialization of Untrusted Data (CWE-502)

2026-03-22 VulDB

GHSA-33x2-ppm4-v46v

Deserialization Pytorch

1.9

CVSS 4.0 · NVD

Severity by source

NVD PRIMARY

1.9 LOW

CVSS:4.0/AV:L/AC:L/AT:N/PR:L/UI:N/VC:L/VI:L/VA:L/SC:N/SI:N/SA:N/E:P/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X

Red Hat

5.3 MEDIUM

qualitative

Primary rating from NVD.

CVSS VectorNVD

CVSS:4.0/AV:L/AC:L/AT:N/PR:L/UI:N/VC:L/VI:L/VA:L/SC:N/SI:N/SA:N/E:P/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X

Attack Vector

Local

Attack Complexity

Low

Privileges Required

Low

User Interaction

None

Scope

Lifecycle Timeline

Severity Changed

Apr 29, 2026 - 01:11 NVD

MEDIUM LOW

CVSS changed

Apr 29, 2026 - 01:11 NVD

5.3 (MEDIUM) 1.9 (LOW)

PoC Detected

Mar 23, 2026 - 14:31 vuln.today

Public exploit code

EUVD ID Assigned

Mar 22, 2026 - 04:30 euvd

EUVD-2026-14280

Analysis Generated

Mar 22, 2026 - 04:30 vuln.today

CVE Published

Mar 22, 2026 - 04:20 nvd

MEDIUM 5.3

Blast Radius

ecosystem impact

† from your stack dependencies † transitive graph · vuln.today resolves 4-path depth

179 pypi packages depend on torch (138 direct, 42 indirect)

Ecosystem-wide dependent count for version 2.10.0.

DescriptionCVE.org

A vulnerability was identified in PyTorch 2.10.0. The affected element is an unknown function of the component pt2 Loading Handler. The manipulation leads to deserialization. The attack can only be performed from a local environment. The exploit is publicly available and might be used. The project was informed of the problem early through a pull request but has not reacted yet.

AnalysisAI

A deserialization vulnerability exists in PyTorch 2.10.0 within the pt2 Loading Handler component, allowing local attackers with low privileges to achieve confidentiality, integrity, and availability impacts through untrusted data processing. The vulnerability (CWE-502) is confirmed to have a publicly available exploit and has been reported to the project via pull request PR#176791, though remediation status remains unclear. With a CVSS score of 5.3 and exploitation probability marked as probable (E:P), this represents a moderate real-world risk primarily affecting local development and deployment environments.

Technical ContextAI

The vulnerability stems from unsafe deserialization practices in PyTorch's pt2 Loading Handler, a component responsible for loading serialized model objects and checkpoint files. CWE-502 (Deserialization of Untrusted Data) is the root cause, which occurs when applications deserialize data from untrusted sources without proper validation or sandboxing. PyTorch users typically interact with this during model loading operations (torch.load(), checkpoint restoration, etc.), making the pt2 Loading Handler a critical attack surface. The affected CPE designation (cpe:2.3:a:n/a:pytorch:*:*:*:*:*:*:*:*) indicates broader PyTorch installations may be vulnerable, though specific version boundaries are currently undefined in available intelligence.

RemediationAI

Immediately check the PyTorch GitHub repository (https://github.com/pytorch/pytorch/) and pull request PR#176791 (https://github.com/pytorch/pytorch/pull/176791) for available patches; upgrade to the patched version once released and thoroughly tested in your environment. In the interim, implement strict access controls limiting local shell access to PyTorch systems, disable model loading from untrusted or user-supplied sources, validate and sandbox all checkpoint and model file sources before deserialization, and monitor for suspicious torch.load() calls in application logs. Additionally, consider running PyTorch workloads in isolated containers or virtual machines with restricted filesystem and network policies to limit blast radius if exploitation occurs.