CVE-2025-4701

The CVE-2025-4701 issue affects VITA-MLLM Freeze-Omni (up to 20250421), specifically the torch.load usage in models/utils.py where improper handling of the path argument enables deserialization and a potential local-host exploit. Root cause: manipulation of the path parameter in torch.load leads ...

PT-2025-21290 · Unknown +1 · Vita-Mllm Freeze-Omni +1

Name of the Vulnerable Software and Affected Versions: VITA-MLLM Freeze-Omni up to 20250421 Description: A problematic issue has been found, affecting the torch.load function in the models/utils.py file. The manipulation of the path argument leads to deserialization, allowing an attack to be...

The vulnerability of the torch.load() function in the PyTorch machine learning framework allows a hacker to execute arbitrary code.

The vulnerability of the torch.load function in the PyTorch machine learning framework is related to deficiencies in the deserialization mechanism. Exploiting this vulnerability could allow a malicious actor to execute arbitrary code remotely...

Deserialization of Untrusted Data

Overview llamafactory is an Easy-to-use LLM fine-tuning framework Affected versions of this package are vulnerable to Deserialization of Untrusted Data through the torch.load function. An attacker can execute arbitrary commands by crafting a malicious .bin file that is then deserialized. PoC pyth...

PYSEC-2025-41

PyTorch is a Python package that provides tensor computation with strong GPU acceleration and deep neural networks built on a tape-based autograd system. In version 2.5.1 and prior, a Remote Command Execution RCE vulnerability exists in PyTorch when loading a model using torch.load with...

CVE-2025-32434 PyTorch: `torch.load` with `weights_only=True` leads to remote code execution

PyTorch is a Python package that provides tensor computation with strong GPU acceleration and deep neural networks built on a tape-based autograd system. In version 2.5.1 and prior, a Remote Command Execution RCE vulnerability exists in PyTorch when loading a model using torch.load with...

Deserialization of Untrusted Data

Overview torch is a Tensors and Dynamic neural networks in Python with strong GPU acceleration Affected versions of this package are vulnerable to Deserialization of Untrusted Data when using the torch.load function on an untrusted model with weightsonly=True, which is documented to be secure. Th...

Deserialization of Untrusted Data

Overview xinference is a Xorbits InferenceXinference is a powerful and versatile library designed to serve language, speech recognition, and multimodal models. With Xorbits Inference, you can effortlessly deploy and serve your or state-of-the-art built-in models using just a single command. Wheth...

CVE-2025-3165 thu-pacman chitu backend.py torch.load deserialization

A vulnerability classified as critical has been found in thu-pacman chitu 0.1.0. This affects the function torch.load of the file chitu/chitu/backend.py. The manipulation of the argument ckptpath/quantckptdir leads to deserialization. An attack has to be approached locally...

PT-2025-14773 · Unknown · Thu-Pacman Chitu

Name of the Vulnerable Software and Affected Versions: thu-pacman chitu version 0.1.0 Description: A critical vulnerability has been found in thu-pacman chitu. This issue affects the torch.load function in the file chitu/chitu/backend.py. The manipulation of the ckpt path/quant ckpt dir argument...

Deserialization of Untrusted Data

Overview vllm is an A high-throughput and memory-efficient inference and serving engine for LLMs Affected versions of this package are vulnerable to Deserialization of Untrusted Data via the hfmodelweightsiterator process due to the usage of the torch.load function with the weightsonly parameter...

PYSEC-2025-58

vLLM is a library for LLM inference and serving. vllm/modelexecutor/weightutils.py implements hfmodelweightsiterator to load the model checkpoint, which is downloaded from huggingface. It uses the torch.load function and the weightsonly parameter defaults to False. When torch.load loads malicious...

PYSEC-2025-58

vLLM is a library for LLM inference and serving. vllm/modelexecutor/weightutils.py implements hfmodelweightsiterator to load the model checkpoint, which is downloaded from huggingface. It uses the torch.load function and the weightsonly parameter defaults to False. When torch.load loads malicious...

PT-2025-5340 · Vllm · Vllm

Name of the Vulnerable Software and Affected Versions: vLLM versions prior to 0.7.0 Description: The issue concerns the vLLM library, specifically the vllm/model executor/weight utils.py file, which implements hf model weights iterator to load model checkpoints downloaded from Hugging Face. It...

Deserialization of Untrusted Data

Overview Affected versions of this package are vulnerable to Deserialization of Untrusted Data via the torch.load function within the Checkpoint.loadcheckpoint method without restrictions. Details Serialization is a process of converting an object into a sequence of bytes which can be persisted t...

Deserialization of Untrusted Data

Overview chattts is an A generative speech model for daily dialogue Affected versions of this package are vulnerable to Deserialization of Untrusted Data when loading the tokenizer with torch.load. An attacker can exploit this by supplying a malicious tokenizer.pt file, leading to code execution...

Remote Code Execution via Unsafe Torch Load in TransfoXLCorpus

Description This is a new bypass to the patch of my previous report, in which the maintainers only apply the "TRUSTREMOTECODE" to guard the vulnerable code of vocabdict = pickle.loadf, but overlooked another vulnerable code of corpusdict = torch.loadresolvedcorpusfile without setting...