9.8
/ 10
CRITICAL
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Description
The mamba language model framework thru 2.2.6 is vulnerable to insecure deserialization (CWE-502) when loading pre-trained models from HuggingFace Hub. The MambaLMHeadModel.from_pretrained() method uses torch.load() to load the pytorch_model.bin weight file without enabling the security-restrictive weights_only=True parameter. This allows the deserialization of arbitrary Python objects via the pickle module. An attacker can exploit this by publishing a malicious model repository on HuggingFace Hub. When a victim loads a model from this repository, arbitrary code is executed on the victim's system in the context of the mamba process.
AI Analysis
Insecure deserialization vulnerability in mamba language model framework
Basic Information
ID
CVE-2026-31239
Source
mitre
Published
May 12, 2026 at 00:00
Modified
May 14, 2026 at 19:54
Affected Product
Vendor
state-spaces
Product
mamba
Version
2.2.6
Affected Versions
n/a n/a n/a
CWE Classification
AI Assessment
AI Score
9.8 / 10
AI Severity
CRITICAL
Vendor
state-spaces
Product
mamba
Version
2.2.6