Out-of-bounds Read Affecting onnx package, versions [,1.22.0)
Threat Intelligence
Snyk has a proof-of-concept or detailed explanation of how to exploit this vulnerability.
The probability is the direct output of the EPSS model, and conveys an overall sense of the threat of exploitation in the wild. The percentile measures the EPSS probability relative to all known EPSS scores. Note: This data is updated daily, relying on the latest available EPSS model version. Check out the EPSS documentation for more details.
Do your applications use this vulnerable package?
In a few clicks we can analyze your entire application and see what components are vulnerable in your application, and suggest you quick fixes.
Test your applications- Snyk IDSNYK-PYTHON-ONNX-17817659
- published5 Jul 2026
- disclosed4 Jul 2026
- creditm00dy
Introduced: 4 Jul 2026
CVE-2026-14647 (opens in a new tab)How to fix?
Upgrade onnx to version 1.22.0 or higher.
Overview
onnx is an Open Neural Network Exchange
Affected versions of this package are vulnerable to Out-of-bounds Read through convPoolShapeInference_opset19 in onnx/defs/nn/old.cc. An attacker can trigger the flaw by supplying a malformed convolution or pooling model that reaches the shape-inference path, causing the runtime to read past the expected tensor shape data during validation. The issue affects deployments that parse untrusted ONNX models and run shape inference on them, including onnxruntime-based services.
Notes
- Affected model shapes are the N- and higher-dimensional convolution/pooling cases; 2D tensors are rejected by the vulnerable path, so the issue is in shape inference for standard spatial operators rather than arbitrary rank-2 inputs.
- The malformed
pads,dilations, andkernel_shapevalues are supplied as operator attributes during validation, so the vulnerable behavior can be reached from static model content without any runtime execution path.