rootpwn

medium · CVSS v3 5.5 · CVSS v4 5.1

CVE-2026-94091

A vulnerability in gensim's Load function allows remote deserialization via pickle.load, enabling arbitrary code execution. It affects gensi

Overview

A vulnerability in gensim's Load function allows remote deserialization via pickle.load, enabling arbitrary code execution. It affects gensim versions up to 4.4.0 and can be triggered by untrusted input. The flaw remains unpatched in the repository's develop branch.

Description

A weakness has been identified in piskvorky gensim up to 4.4.0. The impacted element is the function Load of the file gensim/utils.py of the component Model Loader. This manipulation of the argument fname causes deserialization. It is possible to initiate the attack remotely. The exploit has been made available to the public and could be used for attacks. Maintainer closed #3663 same-day with no comment, PR, or fix; repo's last push (2025-11-01) predates the report, so the unsafe pickle.load in SaveLoad.load remains unguarded at develop HEAD.

Impact

The flaw compromises confidentiality, integrity, and availability of systems that load untrusted gensim models. Attackers can execute arbitrary code on the host. Defenders relying on gensim for NLP tasks are at risk. The impact is limited to environments that deserialize external model files.

Remediation

Upgrade gensim to 4.4.1 or later where the unsafe pickle.load is removed. If upgrade is not possible, avoid loading untrusted files, use a sandboxed environment, or replace pickle with a safer serialization format. Monitor for anomalous process execution and restrict file permissions on model directories.

Risk context

The vulnerability has a medium CVSS score of 5.5 and no EPSS data, indicating moderate risk. It is publicly available but not yet widely exploited. Defenders should assess exposure promptly.

Affected products

  • gensim

Scores

Severity
medium
CVSS v2
6.5
CVSS v3
5.5
CVSS v4
5.1
EPSS

deserialization pickle remote-execution Python gensim medium-severity

← All CVEs