# CVE-2026-12481: Deserialization of Untrusted Data in keras-team/keras

> Attackers can execute arbitrary code via unsafe deserialization in Keras Lambda layers when safe mode is not explicitly enabled.

- **Severity:** High 8.8 (`CVSS:3.0/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H`)
- **Exploited in the wild:** Not known
- **Login needed:** Not needed
- **Attackable:** Over the internet
- **Victim action needed:** A user has to click or open something
- **Fix available:** Yes
- **Published:** 3 Jul 2026
- **Updated:** 6 Jul 2026
- **Also known as:** DEBIAN-CVE-2026-12481, GHSA-5gwj-m78q-7pq3, MINI-9cp3-3crh-vwr4, PYSEC-2026-3631, UBUNTU-CVE-2026-12481

## What is CVE-2026-12481?

CVE-2026-12481 is a high-severity unsafe deserialization vulnerability in keras-team/keras, published on 3 Jul 2026. An attacker can exploit it over the internet without logging in, and only if a user opens a file or follows a link.

### Official description

A vulnerability in keras-team/keras version 3.14.0 allows for arbitrary code execution due to improper handling of deserialization in the `Lambda` layer. Specifically, the `_raise_for_lambda_deserialization()` function fails to enforce the safe-mode guard when `safe_mode` is set to `None`, which is the default value when `from_config()` is called outside of a `SafeModeScope` context. This logic error conflates `None` (unset/default-deny) with `False` (explicitly disabled), bypassing the guard and allowing attacker-controlled `marshal` bytecode to be deserialized. Affected call sites include `keras.layers.deserialize(config)`, `keras.models.clone_model(model)`, and any direct invocation of `Lambda.from_config(config)` without an enclosing `SafeModeScope(True)`. This vulnerability can be exploited to achieve arbitrary OS-level code execution in the context of the server or user process.

Weakness: CWE-502 (Deserialization of Untrusted Data)

## Which versions are affected by CVE-2026-12481?

5 open source packages ship an affected version, including keras. It affects 1 product.


| Package | Ecosystem | Affected | Fixed in |
| --- | --- | --- | --- |
| `keras` | PyPI | All versions | 3.12.3, 3.15.0 |
| `keras` | PyPI | All versions | 3.12.3 |
| `keras` | PyPI | All versions | 3.12.3 |
| `keras` | PyPI | From 3.13.0 | 3.15.0 |
| `keras` | PyPI | From 3.13.0 | 3.15.0 |

Products: Keras Keras.

## How do I fix CVE-2026-12481?

Upgrade keras to a fixed version for your release line: 3.12.3 or 3.15.0.

## Is CVE-2026-12481 being exploited?

It is not confirmed as exploited in the wild, but public exploit code for CVE-2026-12481 exists.

## Other vulnerabilities in keras-team/keras

- [CVE-2024-3660](https://insights.sbomify.com/CVE-2024-3660/): Arbitrary code injection vulnerability in Keras framework < 2.13 (critical 9.8)
- [CVE-2026-1462](https://insights.sbomify.com/CVE-2026-1462/): Safe Mode Bypass in keras-team/keras (high 8.8)
- [CVE-2025-9906](https://insights.sbomify.com/CVE-2025-9906/): Arbitrary Code execution in Keras Safe Mode (high 8.6)
- [CVE-2025-8747](https://insights.sbomify.com/CVE-2025-8747/): Keras safe_mode bypass allows arbitrary code execution when loading a malicious model. (high 8.6)
- [CVE-2026-11816](https://insights.sbomify.com/CVE-2026-11816/): Path Traversal in keras-team/keras (high 8.1)

## Fix prompt for AI coding agents

```text
Check this codebase for CVE-2026-12481 and fix it if it is affected.

What it is: CVE-2026-12481 is a high-severity unsafe deserialization vulnerability in keras-team/keras, published on 3 Jul 2026. An attacker can exploit it over the internet without logging in, and only if a user opens a file or follows a link.
Exploitation: It is not confirmed as exploited in the wild, but public exploit code for CVE-2026-12481 exists.

Affected packages and fixed versions:
- keras (PyPI): affected all versions; fixed in 3.12.3, 3.15.0
- keras (PyPI): affected all versions; fixed in 3.12.3
- keras (PyPI): affected all versions; fixed in 3.12.3
- keras (PyPI): affected from 3.13.0; fixed in 3.15.0
- keras (PyPI): affected from 3.13.0; fixed in 3.15.0

Task:
1. Search the dependency manifests and lockfiles (for example package.json, package-lock.json, requirements.txt, poetry.lock, uv.lock, pom.xml, build.gradle, go.mod, Cargo.lock, Gemfile.lock) and any Dockerfiles for the affected packages, including transitive dependencies.
2. If an affected version is in use, upgrade it to the fixed version for its release line with the project's package manager, so the lockfile is regenerated. Prefer the smallest upgrade that fixes it.
3. If there is no fix, or the upgrade would break the build, do not guess: explain the risk and the safest workaround instead.
4. Run the tests and summarise what changed and why.
5. If nothing in the project is affected, say so and change nothing.

Reference: https://insights.sbomify.com/CVE-2026-12481/ (Markdown: https://insights.sbomify.com/CVE-2026-12481/index.md)
```

## Sources

- CVE record: https://www.cve.org/CVERecord?id=CVE-2026-12481
- NVD: https://nvd.nist.gov/vuln/detail/CVE-2026-12481
- OSV: https://osv.dev/vulnerability/CVE-2026-12481

---
From sbomify Insights: https://insights.sbomify.com/CVE-2026-12481/
Summaries are generated and may contain mistakes; the official description above is the record.