CVE-2026-23001-Hugging-Face-Transformers-Model-Deserialization-Arbitrary-Code-via-Pickle-

CVE-2026-23001 – RCE por deserialización de modelos Hugging Face Transformers Resumen Los modelos de Hugging Face guardados con torch.save utilizan la serialización pickle de Python, que es inherentemente insegura. Si un usuario descarga un modelo aparentemente legítimo de un repositorio, cargarl...

CVE-2025-32434

CVE-2025-32434：PyTorch远程代码执行漏洞 - PoC 什么是CVE-2025-32434？ CVE-2025-32434 是PyTorch中的一个严重远程代码执行（RCE）漏洞，PyTorch是深度学习和神经网络中广泛使用的Python库。 核心问题 该漏洞存在于torch.load函数中，即使使用了看似安全的weightsonly=True参数。weightsonly=True选项旨在通过仅允许加载模型参数来防止代码执行。然而，研究人员找到了一种绕过此保护的方法。 受影响版本 状态| 版本 ---|--- 有漏洞| PyTorch 2.5.1及更早版本 已修复|...

CVE-2025-32434-exploit

概述 此漏洞利用程序展示了 PyTorch 版本低于 2.6.0 中存在的一个严重远程代码执行漏洞（CVE-2025-32434）。该缺陷存在于 torch.load 函数中，即使使用了本应阻止模型反序列化过程中任意代码执行的 weightsonly=True 参数，仍然无法避免。 该漏洞的产生原因是 weightsonlyunpickler 实现未能正确限制反序列化，攻击者可以构造恶意的 .pt 模型文件，在易受攻击的 PyTorch 安装加载该文件时执行任意系统命令。 受影响版本：PyTorch /tmp/whoami" Test the exploit loads the mode...

CVE-2025-50461

CVE-2025-50461: Ejecución Remota de Código mediante Deserialización Insegura de Modelos en Verl Estado: RESERVADO Este ID de CVE ha sido asignado pero aún no está publicado en la lista oficial de CVE. Existe una vulnerabilidad de deserialización en Volcengine's verl 3.0.0, específicamente en el...

CVE-2026-68771_exploit

CVE-2026-68771 — RCE no autenticado por deserialización de pickle en ComfyUI Ejecución remota de código no autenticada en ComfyUI a través del nodo LoadTrainingDataset. LoadTrainingDataset.execute carga los shards del dataset desde el directorio de salida del servidor con torch.loadf y...

CVE-2026-62997

Kedro-Datasets provides data connectors for Kedro. From version 5.0.0 until 9.5.0, kedrodatasetsexperimental.pytorch.PyTorchDataset in kedro-datasets loads .pt model files with torch.load without enforcing weightsonly=True, and user-supplied loadargs are silently dropped. On PyTorch versions...

CVE-2026-62997

Kedro-Datasets (data connectors for the Kedro framework) contains a remote code execution vulnerability in its opt-in kedro_datasets_experimental component, specifically the PyTorchDataset class. From version 5.0.0 through 9.5.0 , .pt model files are loaded via torch.load without enforcing weight...

CVE-2026-62997 Kedro-Datasets: Remote code execution in experimental `PyTorchDataset` via unsafe `torch.load`

Kedro-Datasets provides data connectors for Kedro. From version 5.0.0 until 9.5.0, kedrodatasetsexperimental.pytorch.PyTorchDataset in kedro-datasets loads .pt model files with torch.load without enforcing weightsonly=True, and user-supplied loadargs are silently dropped. On PyTorch versions...

EUVD-2026-81264

Kedro-Datasets provides data connectors for Kedro. From version 5.0.0 until 9.5.0, kedrodatasetsexperimental.pytorch.PyTorchDataset in kedro-datasets loads .pt model files with torch.load without enforcing weightsonly=True, and user-supplied loadargs are silently dropped. On PyTorch versions...

CVE-2026-62997 Kedro-Datasets: Remote code execution in experimental `PyTorchDataset` via unsafe `torch.load`

Kedro-Datasets provides data connectors for Kedro. From version 5.0.0 until 9.5.0, kedrodatasetsexperimental.pytorch.PyTorchDataset in kedro-datasets loads .pt model files with torch.load without enforcing weightsonly=True, and user-supplied loadargs are silently dropped. On PyTorch versions...

CVE-2026-62997 Kedro-Datasets: Remote code execution in experimental `PyTorchDataset` via unsafe `torch.load`

Kedro-Datasets provides data connectors for Kedro. From version 5.0.0 until 9.5.0, kedrodatasetsexperimental.pytorch.PyTorchDataset in kedro-datasets loads .pt model files with torch.load without enforcing weightsonly=True, and user-supplied loadargs are silently dropped. On PyTorch versions...

CVE-2026-90777

ESPnet before 202609 deserializes pretrained model checkpoints using torch.load with weightsonly=False, allowing arbitrary code execution from attacker-supplied files. Attackers can craft malicious checkpoint files that execute code during deserialization when loaded through the initialization or...

CVE-2026-90777 ESPnet before 202609 Remote Code Execution via Unsafe Deserialization

ESPnet before 202609 deserializes pretrained model checkpoints using torch.load with weightsonly=False, allowing arbitrary code execution from attacker-supplied files. Attackers can craft malicious checkpoint files that execute code during deserialization when loaded through the initialization or...

EUVD-2026-77001

ESPnet before 202609 deserializes pretrained model checkpoints using torch.load with weightsonly=False, allowing arbitrary code execution from attacker-supplied files. Attackers can craft malicious checkpoint files that execute code during deserialization when loaded through the initialization or...

CVE-2026-90777 ESPnet before 202609 Remote Code Execution via Unsafe Deserialization

ESPnet before 202609 deserializes pretrained model checkpoints using torch.load with weightsonly=False, allowing arbitrary code execution from attacker-supplied files. Attackers can craft malicious checkpoint files that execute code during deserialization when loaded through the initialization or...

CVE-2026-90777

ESPnet before v.202609 deserializes pretrained model checkpoints using torch.load with weights_only=False, enabling arbitrary code execution from attacker-supplied files. The vulnerable code path is triggered when a malicious checkpoint file is loaded through the model initialization or fine-tuni...

CVE-2026-90777 ESPnet before 202609 Remote Code Execution via Unsafe Deserialization

ESPnet before 202609 deserializes pretrained model checkpoints using torch.load with weightsonly=False, allowing arbitrary code execution from attacker-supplied files. Attackers can craft malicious checkpoint files that execute code during deserialization when loaded through the initialization or...

CVE-2026-90777: Deserialization of Untrusted Data

ESPnet before 202609 deserializes pretrained model checkpoints using torch.load with weightsonly=False, allowing arbitrary code execution from attacker-supplied files. Attackers can craft malicious checkpoint files that execute code during deserialization when loaded through the initialization or...

PT-2026-90848

ESPnet before 202609 deserializes pretrained model checkpoints using torch.load with weights only=False, allowing arbitrary code execution from attacker-supplied files. Attackers can craft malicious checkpoint files that execute code during deserialization when loaded through the initialization o...

PYSEC-2026-3972 PyTorch Lightning load_from_checkpoint has an insecure checkpoint deserialization

PyTorch-Lightning versions 2.6.0 and earlier contain an insecure deserialization vulnerability CWE-502 in the checkpoint loading mechanism. The LightningModule.loadfromcheckpoint method, which is commonly used to load saved model states, internally calls torch.load without setting the...