CVE Feed

    Dashboard / CVE / CVE-2025-9905

    CVE-2025-9905

    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 .h5/.hdf5 model archive that, when loaded via Model.load_model, will trigger arbitrary code to be executed. This is achieved by crafting a special .h5 archive file that uses the Lambda layer feature of keras which allows arbitrary Python code in the form of pickled code. The vulnerability comes from the fact that the safe_mode=True option is not honored when reading .h5 archives. Note that the .h5/.hdf5 format is a legacy format supported by Keras 3 for backwards compatibility.

    Published:Sep 19, 2025
    Last Modified:Feb 26, 2026
    EPS:Sep 19, 2025
    EPSS Score:0.00005
    CVSS Score:7.3

    Affected Products

    Vendor
    Keras
    Product
    Keras

    Exploits

    https://github.com/keras-team/keras/security/advisories/GHSA-36rr-ww3j-vrjv

    Common Attack Pattern Enumeration and Classification (CAPEC)

    No CAPEC recorded yet

    Related CVEs

    CVE-2026-1462 (CVSS 7.8) — A vulnerability in the `TFSMLayer` class of the `keras` package, version 3.13.0, allows attacker-controlled Te…
    CVE-2026-1669 (CVSS 7.1) — Arbitrary file read in the model loading mechanism (HDF5 integration) in Keras versions 3.0.0 through 3.13.1 o…
    CVE-2026-0897 (CVSS 7.1) — Allocation of Resources Without Limits or Throttling in the HDF5 weight loading component in Google Keras 3.0.…
    CVE-2025-12638 (CVSS 8) — Keras version 3.11.3 is affected by a path traversal vulnerability in the keras.utils.get_file() function when…
    CVE-2025-12060 (CVSS 8.9) — The keras.utils.get_file API in Keras, when used with the extract=True option for tar archives, is vulnerable …
    CVE-2025-12058 (CVSS 5.9) — The Keras.Model.load_model method, including when executed with the intended security mitigation safe_mode=Tru…
    CVE-2025-49655 (CVSS 9.8) — Deserialization of untrusted data can occur in versions of the Keras framework running versions 3.11.0 up to b…
    CVE-2025-9906 (CVSS 8.6) — The Keras Model.load_model method can be exploited to achieve arbitrary code execution, even with safe_mode=Tr…
    CVE-2025-8747 (CVSS 8.6) — A safe mode bypass vulnerability in the `Model.load_model` method in Keras versions 3.0.0 through 3.10.0 allow…
    CVE-2025-1550 (CVSS 7.3) — The Keras Model.load_model function permits arbitrary code execution, even with safe_mode=True, through a manu…

    Common Vulnerability Scoring System

    Attack Vector
    Network
    Adjacent
    Local
    Physical
    Privileges Required
    None
    Low
    High
    User Interaction
    None
    Required
    Scope
    Unchanged
    Changed
    Confidentiality
    None
    Low
    High
    Integrity
    None
    Low
    High
    Availability
    None
    Low
    High