---
id: CVE-2026-12484
aliases:
  - GHSA-v2w2-w228-c444
title: 'Keras: TorchModuleWrapper can deserialize unsafe PyTorch pickle data'
summary: 'Keras: TorchModuleWrapper can deserialize unsafe PyTorch pickle data'
severity: high
cvss: 7.8
cvssVector: 'CVSS:3.0/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H'
vendor: keras
product: keras
ecosystem: pip
affected:
  - keras < 3.12.3
  - 'keras >= 3.13.0, < 3.15.0'
patched:
  - keras 3.12.3
  - keras 3.15.0
published: '2026-07-19'
updated: '2026-08-07'
source: OSV
sourceUrl: 'https://osv.dev/vulnerability/GHSA-v2w2-w228-c444'
references:
  - url: 'https://nvd.nist.gov/vuln/detail/CVE-2026-12484'
  - url: 'https://github.com/keras-team/keras/pull/23048'
  - url: 'https://github.com/keras-team/keras/pull/23165'
  - url: >-
      https://github.com/keras-team/keras/commit/55888d3becbdbb45dc16a55b489900f911e2dde5
  - url: >-
      https://github.com/keras-team/keras/commit/d338a45204bdc787c8b3c4a9b82c1911cd52dedf
  - url: 'https://github.com/keras-team/keras'
  - url: 'https://github.com/keras-team/keras/releases/tag/v3.12.3'
  - url: 'https://github.com/keras-team/keras/releases/tag/v3.15.0'
  - url: 'https://huntr.com/bounties/ab14df49-13b5-4442-b754-3189430bfa28'
tags:
  - osv
  - pip
epss: 0.00393
epssPercentile: 0.33333
ingestedAt: '2026-08-08T19:14:54.787Z'
---

## Overview

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.

## Affected packages

- `keras < 3.12.3`
- `keras >= 3.13.0, < 3.15.0`

## Remediation

Upgrade to a patched release:

- `keras 3.12.3`
- `keras 3.15.0`
