Sign in
Deutsch English Español Polski
Critical

PyTorch: `torch.load` with `weights_only=True` leads to remote code execution

  • Disclosed: 18.04.2025
  • Updated: 11.10.2026 14:19

Description

I found a Remote Command Execution (RCE) vulnerability in PyTorch. When loading model using torch.load with weights_only=True, it can still achieve RCE.

Background knowledge

https://github.com/pytorch/pytorch/security As you can see, the PyTorch official documentation considers using torch.load() with weights_only=True to be safe. image Since everyone knows that weights_only=False is unsafe, so they will use the weights_only=True to mitigate the seucirty issue. But now, I just proved that even if you use weights_only=True, it can still achieve RCE.

Credit

This vulnerability was found by Ji'an Zhou.

Affected software

Click a technology to see all vulnerabilities affecting it.

PlatformPackageVulnerable versionsFixed in
Python (PyPI) torch < 2.6.0 2.6.0

References

CVSS score -
CVSS vector
CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N
Weakness (CWE)
CWE-502
Platform
Python (PyPI)
Package
torch
Disclosed
18.04.2025
Updated
11.10.2026 14:19

Sources and licenses

OSV / GitHub Advisory Database - PYSEC-2025-41 - PyPA Advisory Database; CC BY 4.0

OSV / GitHub Advisory Database - GHSA-53q9-r3pm-6pq6 - CC BY 4.0

Vulnerability history: torch

All published vulnerabilities of this package (Python (PyPI)), newest first. Shows whether this is a one-off slip or a package that needs watching.

All (31) in the archive
Total entries
31
First → latest
11.2022 03.2026
Without a fix
9
Exploited in attacks
0

Severity breakdown

  • Critical 3
  • High 10
  • Medium 11
  • Low 5
  • Unrated 2

Disclosures per year

Severity:
  1. 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 pe
    High · 7.8 CVE-2026-4538 <= 2.10.0 no fix
  2. PyTorch is a Python package that provides tensor computation. Prior to version 2.10.0, a vulnerability in PyTorch's `weights_only` unpickler allows an attacker to craft a malicious checkpoint file (`.
    High · 8.8 CVE-2026-24747 < 2.10.0 2.10.0
  3. An issue in pytorch v2.7.0 can lead to a Denial of Service (DoS) when a PyTorch model consists of torch.Tensor.to_sparse() and torch.Tensor.to_dense() and is compiled by Inductor.
    High · 7.5 CVE-2025-55560 < 2.7.1 2.7.1
  4. A buffer overflow occurs in pytorch v2.7.0 when a PyTorch model consists of torch.nn.Conv2d, torch.nn.functional.hardshrink, and torch.Tensor.view-torch.mv() and is compiled by Inductor, leading to a
    High · 7.5 CVE-2025-55558 < 2.7.1 2.7.1
  5. A Name Error occurs in pytorch v2.7.0 when a PyTorch model consists of torch.cummin and is compiled by Inductor, leading to a Denial of Service (DoS).
    High · 7.5 CVE-2025-55557 < 2.7.1 2.7.1
  6. A syntax error in the component proxy_tensor.py of pytorch v2.7.0 allows attackers to cause a Denial of Service (DoS).
    High · 7.5 CVE-2025-55553 < 2.7.1 2.7.1
  7. pytorch v2.8.0 was discovered to display unexpected behavior when the components torch.rot90 and torch.randn_like are used together.
    High · 7.5 CVE-2025-55552 < 2.9.0 2.9.0
  8. An issue in the component torch.linalg.lu of pytorch v2.8.0 allows attackers to cause a Denial of Service (DoS) when performing a slice operation.
    High · 7.5 CVE-2025-55551 < 2.9.0 2.9.0
  9. PyTorch: `torch.load` with `weights_only=True` leads to remote code execution this entry
    Critical CVE-2025-32434 < 2.6.0 2.6.0
  10. A vulnerability was found in PyTorch 2.6.0+cu124. It has been declared as critical. Affected by this vulnerability is the function torch.ops.profiler._call_end_callbacks_on_jit_fut of the component Tu
    High · 7.5 CVE-2025-2148 <= 2.6.0-cu124 no fix
  11. In PyTorch <=2.4.1, the RemoteModule has Deserialization RCE. NOTE: this is disputed by multiple parties because this is intended behavior in PyTorch distributed computing.
    Critical · 9.8 CVE-2024-48063 < 2.5.0 2.5.0
  12. Pytorch before version v2.2.0 was discovered to contain a use-after-free vulnerability in torch/csrc/jit/mobile/interpreter.cpp.
    High · 7.8 CVE-2024-31583 < 2.2.0 2.2.0
  13. In PyTorch before trunk/89695, torch.jit.annotations.parse_type_line can cause arbitrary code execution because eval is used unsafely.
    Critical · 9.8 CVE-2022-45907 < 1.13.1 1.13.1

Know which vulnerabilities affect your application

With the DockRay integration and software inventory enabled, DockRay checks reported component versions against known vulnerabilities and notifies you about confirmed matches. Coverage depends on the supported technology and available advisory data.

Connect your application
DockRay newsletter

Application security in your inbox

Important vulnerabilities, actionable guidance and practical advice for your applications. Choose your technologies and follow what matters to your project.

  • A daily digest of new vulnerabilities, critical ones included.
  • Notifications about important updates.
  • Practical advice on application security and monitoring.

Leave empty to receive updates for every technology.

Privacy policy

Free to subscribe. Unsubscribe with one click in any email.