PYSEC-2026-406 mamba language model framework vulnerable to insecure deserialization when loading pre-trained models from HuggingFace Hub

The mamba language model framework thru 2.2.6 is vulnerable to insecure deserialization CWE-502 when loading pre-trained models from HuggingFace Hub. The MambaLMHeadModel.frompretrained method uses torch.load to load the pytorchmodel.bin weight file without enabling the security-restrictive...

CVE-2026-5843

The MLX inference backend in Docker Model Runner on macOS uses the MLX-LM library, which unconditionally imports and executes arbitrary Python files from model directories via the modelfile configuration field in config.json. When a model's config.json specifies a modelfile pointing to a Python...

CVE-2026-5843

The MLX inference backend in Docker Model Runner on macOS uses the MLX-LM library, which unconditionally imports and executes arbitrary Python files from model directories via the modelfile configuration field in config.json. When a model's config.json specifies a modelfile pointing to a Python...

CVE-2026-5843

The CVE describes a vulnerability in Docker Model Runner (macOS) where the MLX-LM backend unconditionally imports and executes arbitrary Python files specified by model_file in a model's config.json via importlib, without a trust_remote_code gate or sandboxing. This enables container-to-host arbi...

CVE-2026-8751

Affected software: h2oai h2o-3 (versions before 7402). The issue lies in the JAR Handler component, specifically the importBinaryModel() function in h2o-core/src/main/java/hex/Model.java, where input manipulation can trigger deserialization. This enables remote exploitation, with the exploit publ...

CVE-2026-31219

The loadmodel function in the neuralmagictraining.py script of the optimate project in commit a6d302f912b481c94370811af6b11402f51d377f 2024-07-21 is vulnerable to insecure deserialization CWE-502. When a user provides a single model file path e.g., .pt or .pth via the --model command-line argumen...

Deserialization of Untrusted Data

Overview adversarial-robustness-toolbox is a Toolbox for adversarial machine learning. Affected versions of this package are vulnerable to Deserialization of Untrusted Data in the model loading process. An attacker can execute arbitrary code by uploading a maliciously crafted model file to an...

EUVD-2026-29503

The loadmodel function in the neuralmagictraining.py script of the optimate project in commit a6d302f912b481c94370811af6b11402f51d377f 2024-07-21 is vulnerable to insecure deserialization CWE-502. When a user provides a single model file path e.g., .pt or .pth via the --model command-line argumen...

CVE-2026-31219

The loadmodel function in the neuralmagictraining.py script of the optimate project in commit a6d302f912b481c94370811af6b11402f51d377f 2024-07-21 is vulnerable to insecure deserialization CWE-502. When a user provides a single model file path e.g., .pt or .pth via the --model command-line argumen...

CVE-2026-31229

The Adversarial Robustness Toolbox ART thru 1.20.1 contains an insecure deserialization vulnerability CWE-502 in its Kubeflow component's model loading functionality. When loading model weights from a file e.g., model.pt during robustness evaluation, the code uses torch.load without the...

CVE-2026-31219

The loadmodel function in the neuralmagictraining.py script of the optimate project in commit a6d302f912b481c94370811af6b11402f51d377f 2024-07-21 is vulnerable to insecure deserialization CWE-502. When a user provides a single model file path e.g., .pt or .pth via the --model command-line argumen...

CVE-2026-31222

The snorkel library thru v0.10.0 contains an insecure deserialization vulnerability CWE-502 in the Trainer.load method of the Trainer class. The method loads model checkpoint files using torch.load without enabling the security-restrictive weightsonly=True parameter. This default behavior allows...

PT-2026-40058

The load model function in the neural magic training.py script of the optimate project in commit a6d302f912b481c94370811af6b11402f51d377f 2024-07-21 is vulnerable to insecure deserialization CWE-502. When a user provides a single model file path e.g., .pt or .pth via the --model command-line...

PT-2026-40116

The Adversarial Robustness Toolbox ART thru 1.20.1 contains an insecure deserialization vulnerability CWE-502 in its Kubeflow component's model loading functionality. When loading model weights from a file e.g., model.pt during robustness evaluation, the code uses torch.load without the...

PT-2026-40125

Name of the Vulnerable Software and Affected Versions Ludwig framework versions prior to 0.10.5 Description The model serving component is subject to insecure deserialization. When initiating a model server via the ludwig serve command, the framework utilizes the torch.load function to load model...

CVE-2026-31219

The connected documents confirm a concrete vulnerability in the optimate project: the _load_model() (or load_model()) function in neural_magic_training.py deserializes a single model file passed via --model using torch.load() without weights_only=True, enabling arbitrary Python object deserializa...

CVE-2026-31229

The Adversarial Robustness Toolbox ART thru 1.20.1 contains an insecure deserialization vulnerability CWE-502 in its Kubeflow component's model loading functionality. When loading model weights from a file e.g., model.pt during robustness evaluation, the code uses torch.load without the...

GHSA-MGX6-5CF9-RR43 Keras vulnerable to DoS via Malicious .keras Model (HDF5 Shape Bomb Causes Petabyte Allocation in KerasFileEditor)

Summary Keras’s model loader KerasFileEditor unsafely loads user-supplied .keras model files containing HDF5-based weight files without performing any validation on HDF5 dataset metadata. An attacker can craft a .keras archive containing a valid model.weights.h5 file whose dataset declares an...

SGLang CVE-2026-5760 (CVSS 9.8) Enables RCE via Malicious GGUF Model Files

A critical security vulnerability has been disclosed in SGLang that, if successfully exploited, could result in remote code execution on susceptible systems. The vulnerability, tracked as CVE-2026-5760 , carries a CVSS score of 9.8 out of 10.0. It has been described as a case of command injection...

Arbitrary Code Injection

Overview sglang is a SGLang is a fast serving framework for large language models and vision language models. Affected versions of this package are vulnerable to Arbitrary Code Injection via the reranking endpoint when a model file containing a malicious tokenizer.chattemplate is loaded, due to...