CVE-2026-6840

CVE-2026-6840 describes missing bounds validation for an operator during model loading, enabling a out-of-range operator-code lookup. Affected versions are those prior to commit 1.30.0. The CVSS 3.1 base score is 5.5 (Medium) with Local attack vector, Low attack complexity, No privileges required...

CVE-2026-6840

Missing bounds validation for operator could allow out of range operator-code lookup during model loading Affected version is prior to commit 1.30.0...

CVE-2026-6840

Missing bounds validation for operator could allow out of range operator-code lookup during model loading Affected version is prior to commit 1.30.0...

CVE-2026-6840

Missing bounds validation for operator could allow out of range operator-code lookup during model loading Affected version is prior to commit 1.30.0...

PT-2026-34263

CVE-2026-6840 Missing bounds validation for operator could allow out of range operator-code lookup during model loading Affected version is prior to commit 1.30.0. https://t.co/DGJUzFs4hC...

ONE 输入验证错误漏洞

ONE is a high-performance edge-side neural network inference framework developed by Samsung. Versions of ONE prior to 1.30.0 contained a vulnerability related to input validation errors. This vulnerability stemmed from a lack of operator boundary validation, which could lead to out-of-bounds code...

Rembg has a Path Traversal via Custom Model Loading

Summary A path traversal vulnerability in the rembg HTTP server allows unauthenticated remote attackers to read arbitrary files from the server's filesystem. By sending a crafted request with a malicious modelpath parameter, an attacker can force the server to attempt loading any file as an ONNX...

CVE-2026-40086

Rembg: Path traversal in the HTTP server allows unauthenticated remote attackers to read arbitrary files via a crafted model_path parameter. Affected versions are prior to 2.0.75; the issue can reveal file existence, permissions, and potentially contents through error messages. The vulnerability ...

Security Bulletin: Arbitrary File Read, SSRF, and Code Execution Vulnerabilities in TensorFlow Keras Model Loading (v2.13) affects watsonx.data

Summary A vulnerability in TensorFlow Keras v2.13 allows malicious .keras model files to trigger arbitrary local file reads, Server-Side Request Forgery SSRF, and potential code execution during model loading—even when safemode=True is enabled. The issue arises from improper handling of external...

CVE-2026-34446

A flaw was found in Open Neural Network Exchange ONNX, an open standard for machine learning interoperability. The onnx.load function, which is used to load machine learning models, does not correctly handle hardlinks. This vulnerability could allow an attacker to create a specially crafted ONNX...

PYSEC-2026-103

Open Neural Network Exchange ONNX is an open standard for machine learning interoperability. In versions up to and including 1.20.1, a security control bypass exists in onnx.hub.load due to improper logic in the repository trust verification mechanism. While the function is designed to warn users...

UBUNTU-CVE-2026-28500

Open Neural Network Exchange ONNX is an open standard for machine learning interoperability. In versions up to and including 1.20.1, a security control bypass exists in onnx.hub.load due to improper logic in the repository trust verification mechanism. While the function is designed to warn users...

CVE-2026-28500 ONNX Untrusted Model Repository Warnings Suppressed by silent=True in onnx.hub.load() — Silent Supply-Chain Attack

Open Neural Network Exchange ONNX is an open standard for machine learning interoperability. In versions up to and including 1.20.1, a security control bypass exists in onnx.hub.load due to improper logic in the repository trust verification mechanism. While the function is designed to warn users...

CVE-2026-28500 ONNX Untrusted Model Repository Warnings Suppressed by silent=True in onnx.hub.load() — Silent Supply-Chain Attack

Open Neural Network Exchange ONNX is an open standard for machine learning interoperability. In versions up to and including 1.20.1, a security control bypass exists in onnx.hub.load due to improper logic in the repository trust verification mechanism. While the function is designed to warn users...

CVE-2026-3071

Deserialization of untrusted data in the LanguageModel class of Flair from versions 0.4.1 to latest are vulnerable to arbitrary code execution when loading a malicious model...

CVE-2026-3071

Deserialization of untrusted data in the LanguageModel class of Flair from versions 0.4.1 to latest are vulnerable to arbitrary code execution when loading a malicious model...

`trust_remote_code=False` Bypass in LightGlue Nested Config Resolution (Transformers 5.2.0) Leading to Remote Code Execution During Normal `from_pretrained()` Loading

Description Transformers contains a trust-boundary flaw in the LightGlue loading path. When loading a LightGlue model, LightGlueConfig reads trustremotecode from untrusted model config.json and reuses it for nested AutoConfig.frompretrained... resolution. This allows an attacker-controlled model...

Arbitrary File Write via Path Traversal in Orbax Checkpoint Asset Dict Keys

Description When loading a Keras model from an Orbax checkpoint directory, the writenesteddicttodir function uses dict keys from the checkpoint's asset data directly in os.path.join without any path sanitization. A crafted Orbax checkpoint can include absolute paths or path traversal sequences .....

Keras Has A Local File Disclosure Via HDF5 External Storage During Keras Weight Loading

Summary TensorFlow / Keras continues to honor HDF5 “external storage” and "ExternalLink" features when loading weights. A malicious ".weights.h5" or a ".keras" archive embedding such weights can direct "loadweights" to read from an arbitrary readable filesystem path. The bytes pulled from that pa...

Deserialization of Untrusted Data

Overview nemo-toolkit is a NeMo - a toolkit for Conversational AI Affected versions of this package are vulnerable to Deserialization of Untrusted Data via the model loading process with weightsonly=False. An attacker can execute arbitrary code, escalate privileges, disclose sensitive information...