OSSForks/pytorch
1
0
Fork 0
mirror of https://github.com/zebrajr/pytorch.git synced 2025-12-06 12:20:52 +01:00
Code Issues Actions 38 Packages Projects Releases Wiki Activity

[Security] Advise against loading untrusted TorchScripts (#152336)

Browse Source 
As torchscripted model is a Turing complete program
Pull Request resolved: https://github.com/pytorch/pytorch/pull/152336
Approved by: https://github.com/albanD

Co-authored-by: albanD <desmaison.alban@gmail.com>
This commit is contained in:
Nikita Shulga 2025-04-28 20:18:53 +00:00 committed by PyTorch MergeBot
parent 00ebbbb701
commit 33766de2d3
1 changed files with 5 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
SECURITY.md
View File

@ -3,6 +3,7 @@
- [**Reporting a Vulnerability**](#reporting-a-vulnerability) - [**Reporting a Vulnerability**](#reporting-a-vulnerability)
- [**Using Pytorch Securely**](#using-pytorch-securely) - [**Using Pytorch Securely**](#using-pytorch-securely)
- [Untrusted models](#untrusted-models) - [Untrusted models](#untrusted-models)
- [TorchScript models](#torchscript-models)
- [Untrusted inputs](#untrusted-inputs) - [Untrusted inputs](#untrusted-inputs)
- [Data privacy](#data-privacy) - [Data privacy](#data-privacy)
- [Using distributed features](#using-distributed-features) - [Using distributed features](#using-distributed-features)
@ -38,6 +39,10 @@ Important Note: The trustworthiness of a model is not binary. You must always de
https://arxiv.org/abs/2312.04748 https://arxiv.org/abs/2312.04748
https://arxiv.org/abs/2401.05566 https://arxiv.org/abs/2401.05566
### TorchScript models
TorchScript models should treated the same way as locally executable code from an unknown source. Only run TorchScript models if you trust the provider. Please note, that tools for introspecting TorchScript models (such as `torch.utils.model_dump`) may also execute partial or full code stored in those models, therefore they should be used only if you trust the provider of the binary you are about to load.
### Untrusted inputs during training and prediction ### Untrusted inputs during training and prediction
If you plan to open your model to untrusted inputs, be aware that inputs can also be used as vectors by malicious agents. To minimize risks, make sure to give your model only the permissions strictly required, and keep your libraries updated with the latest security patches. If you plan to open your model to untrusted inputs, be aware that inputs can also be used as vectors by malicious agents. To minimize risks, make sure to give your model only the permissions strictly required, and keep your libraries updated with the latest security patches.