Browse / AI Framework and Platform Flaws / CVE-2025-2149
PyTorch: Manipulation of the argument scale/zero_point leads to improper initialization via Quantized Sigmoid Module
- Class
- AI Framework and Platform Flaws (filed by a model that read the advisory)
- Technique
- Other weaknesses
- Affected
- torch (PyPI)
- CVSS
- CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:N/I:L/A:N
- EPSS
- 0.3% chance of exploitation in 30 days
How it works, in short
The quantized sigmoid in PyTorch leaves its output state improperly initialised when scale or zero_point arguments are manipulated. (Our summary, written by a model. The source text is below.)
Details
A vulnerability was found in PyTorch 2.6.0+cu124. It has been rated as problematic. Affected by this issue is the function nnq_Sigmoid of the component Quantized Sigmoid Module. The manipulation of the argument scale/zero_point leads to improper initialization. The attack needs to be approached locally. The complexity of an attack is rather high. The exploitation is known to be difficult. The exploit has been disclosed to the public and may be used.
References
- https://nvd.nist.gov/vuln/detail/CVE-2025-2149
- https://github.com/pytorch/pytorch/issues/147818
- https://github.com/pytorch/pytorch/issues/147818#issue-2877301660
- https://github.com/pypa/advisory-database/tree/main/vulns/torch/PYSEC-2025-190.yaml
- https://github.com/pytorch/pytorch
- https://vuldb.com/?ctiid.299060
- https://vuldb.com/?id.299060
- https://vuldb.com/?submit.506563
Source: GitHub Advisory Database, used under CC BY 4.0.