CVE-2026-15976

SGLang contains a Remote Code Execution (RCE) vulnerability when loading model weights from a HuggingFace repository via /update_weights_from_disk. The issue arises because torch.load(..., weights_only=False) can fall back to pickle deserialization of .bin files, enabling arbitrary code execution...

CVE-2026-15976 CVE-2026-15976

SGLang contains a RCE vulnerability when attempting to load model weights from a HuggingFace repository, specifically within the /updateweightsfromdisk, where torch.load..., weightsonly=False fallback enables pickle deserialization of .bin files...

EUVD-2026-51264

SGLang contains a RCE vulnerability when attempting to load model weights from a HuggingFace repository, specifically within the /updateweightsfromdisk, where torch.load..., weightsonly=False fallback enables pickle deserialization of .bin files...

PT-2026-66561

SGLang contains a RCE vulnerability when attempting to load model weights from a HuggingFace repository, specifically within the /update weights from disk, where torch.load..., weights only=False fallback enables pickle deserialization of .bin files...

PYSEC-2026-2852 OpenMed vulnerable to remote code injection through privacy-filter model loading path

OpenMed before 1.5.2 contains a remote code execution vulnerability in the PII privacy-filter model loading path. The privacy-filter dispatcher used broad substring matching on the user-supplied modelname parameter, allowing a value such as attacker/foo-privacy-filter-bar to route through a path...

PYSEC-2026-3062 SGLang has an Improper Input Validation/Injection Issue

A vulnerability was detected in sgl-project SGLang up to 0.5.9. Impacted is the function gettokenizer of the file python/sglang/srt/utils/hftransformersutils.py of the component HuggingFace Transformer Handler. The manipulation results in deserialization. The attack can be executed remotely. A hi...

PYSEC-2026-2522 InstructLab Includes Functionality from Untrusted Control Sphere

A flaw was found in InstructLab. The linuxtrain.py script hardcodes trustremotecode=True when loading models from HuggingFace. This allows a remote attacker to achieve arbitrary Python code execution by convincing a user to run ilab train/download/generate with a specially crafted malicious model...

CVE-2026-54316

A flaw was found in Claude Code, an agentic coding tool. An attacker could exploit a misconfiguration in the tool's web request functionality, known as WebFetch, where the huggingface.co domain was pre-approved without proper path restrictions. By injecting untrusted content into a Claude Code...

PYSEC-2026-1415 gradio vulnerable to Path Traversal

An issue was discovered in gradio-app/gradio, where the /componentserver endpoint improperly allows the invocation of any method on a Component class with attacker-controlled arguments. Specifically, by exploiting the moveresourcetoblockcache method of the Block class, an attacker can copy any fi...

Malicious code in zod-pino434 (npm)

--- -= Per source details. Do not edit below this line.=- Source: amazon-inspector 599a5d533bf699156b78f5296f643227bf3e43d8e46f2adabfe250205a05bd26 Package name zod-pino434 and package.json description Node.js integration layer for Autodesk Forge do not match the shipped code. On npm install, the...

Malicious code in zod-pino444 (npm)

--- -= Per source details. Do not edit below this line.=- Source: amazon-inspector 560c3d47344d4da5dffac230236b5248d2f1856c221ffac7ff72d4e3b197957b Package [email protected] is a credential/secret harvester disguised behind a benign-sounding name. scripts/postinstall-agent.mjs is a...

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...

PYSEC-2026-542 smolagents has Sandbox Escape Vulnerability in the local_python_executor.py Module

A sandbox escape vulnerability was identified in huggingface/smolagents version 1.14.0, allowing attackers to bypass the restricted execution environment and achieve remote code execution RCE. The vulnerability stems from the localpythonexecutor.py module, which inadequately restricts Python code...

CVE-2026-41523

A flaw was found in vLLM, an inference and serving engine for large language models LLMs. An unauthenticated attacker can exploit an assert-based security check during activation function loading. By publishing a malicious HuggingFace model, an attacker can achieve arbitrary code execution on the...

Out Of Band Data Exfiltration

Claude Code is vulnerable to Out-of-Band Data Exfiltration. The vulnerability is due to the pre-approval of the hostname huggingface.co as a bare hostname for the WebFetch tool, where any path on that domain—including attacker-controlled model repositories—was auto-approved without a permission...

CVE-2026-54316

CVE-2026-54316 (Claude Code) affects Claude Code versions 0.2.54–2.1.162, fixed in 2.1.163. The WebFetch tool allowed any path on the pre-approved domain huggingface.co, enabling an attacker-controlled repository path to be fetched without prompts or --allowedTools restrictions. If an attacker ca...

CVE-2026-54316

Claude Code is an agentic coding tool. From 0.2.54 until 2.1.163, because the hostname huggingface.co was pre-approved as a bare hostname for the WebFetch tool, any path on that domain—including attacker-controlled model repositories—was auto-approved without a permission prompt or being subject ...

CVE-2026-54316 Claude Code: Out-of-Band Data Exfiltration via Pre-Approved HuggingFace Domain in WebFetch

Claude Code is an agentic coding tool. From 0.2.54 until 2.1.163, because the hostname huggingface.co was pre-approved as a bare hostname for the WebFetch tool, any path on that domain—including attacker-controlled model repositories—was auto-approved without a permission prompt or being subject ...

CVE-2026-54316 Claude Code: Out-of-Band Data Exfiltration via Pre-Approved HuggingFace Domain in WebFetch

Claude Code is an agentic coding tool. From 0.2.54 until 2.1.163, because the hostname huggingface.co was pre-approved as a bare hostname for the WebFetch tool, any path on that domain—including attacker-controlled model repositories—was auto-approved without a permission prompt or being subject ...

CVE-2026-41523

vLLM is an inference and serving engine for large language models LLMs. Prior to 0.22.0, an assert-based security check in vLLM's activation function loading allows any unauthenticated attacker to achieve arbitrary code execution on the server by publishing a malicious HuggingFace model, when vLL...