CVE-2026-31239 | The mamba language model framework thru 2.2.6 is vulnerable … | CVSS 9.8 CRITICAL

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.

Metadata

CVE ID: CVE-2026-31239
State: PUBLISHED
Assigner: mitre
Reserved: 2026-03-09 00:00 UTC
Published: 2026-05-12 00:00 UTC
Last updated: 2026-05-14 19:54 UTC
Primary CWE: CWE-502
CWE-502 Deserialization of Untrusted Data
Vendor / Product: n/a / n/a
Sources: cve.org · NVD

Severity & Metrics

9.8 CRITICAL CVSS 3.1

CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

SSVC — CISA Coordinator

Exploitation: none
Automatable: yes
Tech. Impact: total

Affected products (1)

Vendor	Product	Platform	Versions
n/a	n/a	—	n/a

Weakness (CWE)

CWE	Source	Description
—	cna	n/a
CWE-502	adp	CWE-502 Deserialization of Untrusted Data

CVSS scores (1)

Score	Severity	Version	Source	Vector
9.8	CRITICAL	3.1	adp	CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

References (2)