CyberSecBulletin
CVE-2025-9906NVD status: Analyzed

Keras: Deserialization of Untrusted Data

Published Updated By the CyberSecBulletin desk

CVE-2025-9906 is a high-severity deserialization of untrusted data vulnerability in Keras from Keras-team, with a CVSS 3.1 base score of 7.3 out of 10. It can be exploited with local access to the system by an attacker with a low-privileged account, but only if a user takes part, for example by opening a link or file. The record lists version 3.11.0 as the first without the flaw, so updating Keras to that release or later removes the exposure. There is no confirmed exploitation so far. EPSS estimates a 0.2% chance of exploitation within 30 days, which is low compared with other CVEs.

Next patch cycle

High severity with no evidence of exploitation so far.

What to do

  1. Update Keras to 3.11.0 or a later version.
  2. Because the attack needs a user to take part, warn users about unexpected links and files related to this product until it is patched.

Affected products and versions

VendorProductAffected versionsFixed in
Keras-teamKeras3.0.0 up to, but not including, 3.11.03.11.0

Version data comes from the organisation that assigned the CVE. Confirm against the vendor advisory before deciding you are unaffected.

How the severity score breaks down

Attack vector
LocalThe attacker needs local access or must trick a user into running something.
Attack complexity
LowNo special conditions are needed; the attack works reliably.
Privileges required
LowA basic user account is enough.
User interaction
RequiredA user must take an action such as opening a file or link.
Scope
UnchangedImpact stays within the vulnerable component.
Confidentiality
HighAll data handled by the component can be read.
Integrity
HighData or code can be modified at will.
Availability
HighThe component can be taken down completely.

Vector: CVSS:3.1/AV:L/AC:L/PR:L/UI:R/S:U/C:H/I:H/A:H. Scored by NVD.

Scoring systemScoreSeverityScored by
CVSS 4.08.6HighCNA (google.com)
CVSS 3.17.3HighNVD

Official 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.

Source: the CVE record as published by its numbering authority, via NVD and CVE.org.

The underlying weakness: Deserialization of Untrusted Data

The product deserializes untrusted data without sufficiently ensuring that the resulting data will be valid.

How developers prevent this class of flaw

  • If available, use the signing/sealing features of the programming language to assure that deserialized data has not been tainted. For example, a hash-based message authentication code (HMAC) could be used to ensure that data has not been modified.
  • When deserializing data, populate a new object rather than just deserializing. The result is that the data flows through safe input validation and that the functions are safe.
  • Explicitly define a final object() to prevent deserialization.

More Deserialization of Untrusted Data vulnerabilities (CWE-502)

Timeline

  1. CVE record published.
  2. CISA published its SSVC assessment (exploitation: none, automatable: no, technical impact: total).
  3. Record last changed materially (score, affected products, status or references).

References

Patches and release notes

Other vulnerabilities in Keras

All Keras vulnerabilities

Recent Deserialization of Untrusted Data vulnerabilities

Spotted an error? Tell us.