Skip to content
OpenAdvisories
High · 8.6 CVE-2025-9906 EUVD-2025-30281GHSA-36fq-jgmw-4r9cPYSEC-2025-76

Arbitrary Code execution in Keras Safe Mode

Published Sep 19, 2025 · Updated Feb 26, 2026 · Assigned by Google

Description

The Keras Model.load_model method can be exploited to achieve arbitrary code execution, even with safe_mode=True.

One can create a specially crafted .keras model archive that, when loaded via Model.load_model, will trigger arbitrary code to be executed. This is achieved by crafting a special config.json (a file within the .keras archive) that will invoke keras.config.enable_unsafe_deserialization() to disable safe mode. Once safe mode is disable, one can use the Lambda layer feature of keras, which allows arbitrary Python code in the form of pickled code. Both can appear in the same archive. Simply the keras.config.enable_unsafe_deserialization() needs to appear first in the archive and the Lambda with arbitrary code needs to be second.

Affected products

VendorProductAffectedFixed
Keras-team Keras 3.0.0 – < 3.11.0 3.11.0

Fixed in Linux distributions

Distributions backport fixes, so the version that fixes this vulnerability differs per release — upgrade to at least the version shown. From the distributions' security trackers and advisories.

ReleasePackageFixed inAdvisory
Debian 11 keras No fix yet —
ReleasePackageFixed inAdvisory
Ubuntu 20.04 LTS keras No fix yet —
Ubuntu 18.04 LTS keras No fix yet —

Azure Linux

ReleasePackageFixed inAdvisory
Azure Linux 3 keras 3.3.3-4 —

Wolfi

ReleasePackageFixed inAdvisory
Wolfi rolling kubeflow-pipelines-visualization-server 2.17.0-r0 —

Affected packages

EcosystemPackageAffectedFixedRecord
PyPI keras < 3.11.0 3.11.0 GHSA-36fq-jgmw-4r9c
PyPI keras — — PYSEC-2025-76

References