Vulnerability · Published Jul 3, 2026 · Assigned by @huntr_ai
CVE-2026-12481Deserialization of Untrusted Data in keras-team/keras
High 8.8
In plain English
A high-severity unsafe deserialization in keras-team/keras from keras-team. Proof-of-concept exploit code is public. The chance it is exploited in the next 30 days is estimated at 0.7% (higher than 54% of all known vulnerabilities). Fixed versions are available - upgrade the affected package.
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.
Affected packages & fixed versions
| Ecosystem | Package | Fixed in | Advisory |
|---|---|---|---|
| PyPI | keras | 3.12.3 | GHSA-5gwj-m78q-7pq3 |
| PyPI | keras | 3.15.0 | GHSA-5gwj-m78q-7pq3 |
| PyPI | keras | 3.12.3, 3.15.0 | PYSEC-2026-3631 |
Affected products (as reported)
| Vendor | Product | Versions |
|---|---|---|
| keras-team | keras-team/keras | unspecified – through latest |
Weakness: CWE-502 · Vector: CVSS:3.0/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H