Advisoriesby IbisPyTorchPYSEC-2024-259
PyTorch: unsafe deserialization
Fix: upgrade to 2.5.0 or later
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.
Affected versions
| Package | Affected | Fixed in |
|---|---|---|
| torch PyPI | < 2.5.0 | 2.5.0 |
Details and references
- CVSS 3.1
- CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
- Severity from
- the CVSS score
- Also known as
- BIT-pytorch-2024-48063, CVE-2024-48063
- rumbling-slice-eb0.notion.site/Distributed-RPC-Framework-RemoteModule-has-Deserialization-RCE-in-pytorch-pytorch-111e3cda9e8c8021a7d3cbc61ee1a20c
- github.com/pytorch/pytorch/issues/129228
- gist.github.com/hexian2001/c046c066895a963ecc0a2cf9e1180065
- github.com/pytorch/pytorch/security/policy#using-distributed-features
More PyTorch advisories
All PyTorch
PyTorch is vulnerable to memory corruption through its unpack_sequence functionCVE-2025-2999fixed in 2.9.1PyTorch is vulnerable to memory corruption through its torch.jit.script functionCVE-2025-3000fixed in 2.13.0PyTorch is Vulnerable to Memory Consumption through pad_packed_sequence FunctionCVE-2025-2998no fix yetPyTorch susceptible to local Denial of ServiceCVE-2025-2953fixed in 2.7.1-rc1A vulnerability was found in PyTorch 2.6.0+cu124CVE-2025-2149no fix yetPyTorch Tuple Handler is Vulnerable to Memory Corruption through Manipulation of None ArgumentCVE-2025-2148no fix yet