ZDF
Sign inSign up

Vulnerabilities

Vulnerability · CVE-2025-9906

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

Published Sep 19, 2025 · catalog updated Oct 19, 2026 · Keras-team

Last checked Oct 26, 2026. A connected source was read within the last 7 days.

Check if you run it

No confirmed exploitation is on record for CVE-2025-9906. EPSS is 0.20%. Act only if a listed version of Keras-team is installed.

Gaps on this record

  • No vendor fix or required action is stored. Compare the linked advisory before you change production.

Same product and reporting

Severity
High8.6
Known exploitation
Not in CISA/VulnCheck KEV
EPSS
0.20%
Remediation
Version ranges on file
Affected products
cpe keras/keras · vendor Keras-team keras 3.0.0 <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.

Are you affected?

Enter the version you run. The result uses only the ranges stored on this record.

The check returns affected, not affected, or incomplete. It does not guess a version that is missing from the record.

  • cpe keras/kerasversion unspecified
  • vendor Keras-team keras 3.0.0 <3.11.03.0.0 <3.11.0

How to fix CVE-2025-9906

What to do

Three questions, each answered from this record only. Confirm the vendor advisory before you patch.

Should you patch?

No confirmed exploitation is on record for CVE-2025-9906. EPSS is 0.20%. Act only if a listed version of Keras-team is installed.

Which versions are affected?

Compare what you have installed with Keras-team keras 3.0.0 <3.11.0 3.0.0 <3.11.0. Compare the installed version with the linked vendor advisory before you change production.

What should an analyst verify?

CWE-502: Deserialization of Untrusted Data. EPSS 0.20%. There is no CISA KEV listing. No Exploit-DB or Metasploit reference is linked. Use the source links for the original advisory.

Code

The project this record applies to. Exploit listings stay in References.

Timeline

Detection, updates, and resolution for this record.

  1. Sep 19, 2025 · NIST NVD

    Detected

    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(

  2. Oct 19, 2026 · NIST NVD

    Updated

    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(

Sources

  • CVE.orgVulnerability recordCVE-2025-9906

    Official CVE identifier

    cve.org
  • FIRST EPSSExploitation probabilityCVE-2025-9906

    30-day exploitation probability 0.20%

    first.org
  • EUVDVulnerability recordEUVD-2025-30281

    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 crafti

    euvd.enisa.europa.eu
  • NIST NVDVulnerability recordCVE-2025-9906

    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 craft

    nvd.nist.gov

References

Freshness

Last verified in this catalog Oct 26, 2026. Significant claims (KEV, EPSS, CVSS, affected ranges) come from the source rows above — open the original publisher for the authoritative advisory.

  • FIRST EPSSOct 23, 2026
  • EUVDOct 22, 2026
  • CVE.orgOct 21, 2026
  • NIST NVDOct 19, 2026