Skip to main content
HIGH

CVE-2026-12484

CVSS v3

7.8

HIGH

EPSS Score

0.4 %

exploit probability, as of 2026-10-17

CISA KEV

No

known exploited

Exploitation

—

SSVC status

Get an alert for each new CVE published for keras. Watch keras in CVE Watch Free account · no credit card

Affected and fixed versions

Packages from OSV.dev and products from the vendors' CSAF advisories.

version not verifiedVersions not verified: read as OSV.dev and the vendors publish them, not checked against each vendor's release list.

Packages

  • kerasPyPIall versions · fixed in 3.12.3

Vendor advisories

  • CVE-2026-12484 Red Hat Product Security4 known affected

    keras: torch: Keras: Arbitrary code execution via unsafe deserialization of PyTorch data

    rhoai/odh-pipeline-runtime-tensorflow-cuda-py312-rhel9 as a component of Red Hat OpenShift AI (RHOAI) · rhoai/odh-pipeline-runtime-tensorflow-rocm-py312-rhel9 as a component of Red Hat OpenShift AI (RHOAI) · rhoai/odh-workbench-jupyter-tensorflow-cuda-py312-rhel9 as a component of Red Hat OpenShift AI (RHOAI)

Full record on NVD

Description

A vulnerability in keras-team/keras version 3.15.0 allows unsafe deserialization of attacker-controlled PyTorch pickle data through the public `keras.layers.TorchModuleWrapper.from_config` method. This method invokes `torch.load(..., weights_only=False)` without requiring an explicit unsafe opt-in, such as a `safe_mode=False` parameter. When called outside a `SafeModeScope(True)` context, the absence of an ambient safe mode state permits unsafe deserialization by default. This issue can lead to arbitrary code execution if untrusted Keras layer configurations are processed using this method. The vulnerability arises because the method does not enforce safe deserialization practices unless explicitly guarded by Keras safe mode.

Technical details

CVSS v3 Vector
CVSS:3.0/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
Published
2026-07-19
Last Modified
2026-07-20

Frequently asked questions

What is CVE-2026-12484?

A vulnerability in keras-team/keras version 3.15.0 allows unsafe deserialization of attacker-controlled PyTorch pickle data through the public `keras.layers.TorchModuleWrapper.from_config` method. This method invokes `torch.load(..., weights_only=False)` without requiring an explicit unsafe opt-in, such as a `safe_mode=False` parameter. When called outside a `SafeModeScope(True)` context, the absence of an ambient safe mode state permits unsafe deserialization by default. This issue can lead to arbitrary code execution if untrusted Keras layer configurations are processed using this method. The vulnerability arises because the method does not enforce safe deserialization practices unless explicitly guarded by Keras safe mode.

Is CVE-2026-12484 actively exploited?

Active exploitation of CVE-2026-12484 has not been confirmed. Its EPSS score was 0.4% on 2026-10-17, the estimated probability of exploitation in the next 30 days.

What is the CVSS score for CVE-2026-12484?

CVE-2026-12484 has a CVSS v3 base score of 7.8 (HIGH severity), with vector string CVSS:3.0/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H.

Running keras? Hear about its next vulnerability

CVE Watch matches each new CVE to the vendors and products you list (vendor and product, not the version) and puts it in your alerts with its KEV and EPSS status.

Get alerts for this productCISA KEV additions, October 2026Search CVEs

Already have an account? Open CVE WatchNo credit card required · 50 free checks/month · Free API key

Go further

Related vulnerabilities

Same product first, then the same vendor advisory and the same publication week.

Back to CVE Database