CVE-2025-9906

published 19 Sep 2025 · updated 26 Feb 2026

Arbitrary Code execution in Keras Safe Mode

8.6HighCVSS 8.6EPSS 0%

No exploitation in the wild has been reported. EPSS puts the chance of exploitation in the next 30 days at 0%. Fixed in 3.11.0.

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

Vendors, products and packages with version ranges

VendorProductVersionsFixed
keras-team>=3.0.0 <3.11.03.11.0

Timeline

Every event on this vulnerability

Embed

A live severity badge for an advisory, README or ticket

CVE-2025-9906 severity on Zeptric[![CVE-2025-9906](https://zeptric.io/badge/cve/CVE-2025-9906.svg)](https://zeptric.io/cve/CVE-2025-9906)