CVE-2026-115920 ONNX: Out-of-bounds read in ConvTranspose shape inference via undersized weight rank
- Exploited in the wild
- Not known
- Fix
- Available
What is CVE-2026-115920?
CVE-2026-115920 is a medium-severity vulnerability in onnx, published on 26 Oct 2026.
Official description
Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. From 1.2.1 until 1.22.0, ConvTranspose shape inference processes a weight tensor with fewer than three dimensions without a kernel_shape attribute by deriving an empty effective_kernel_shape and indexing it for the input's spatial dimensions. The flaw affects convTransposeShapeInference_opset11() and the current inference path, and an untrusted model that can pass onnx.checker.check_model() can crash onnx.shape_inference.infer_shapes(), causing denial of service without demonstrated confidentiality or integrity impact. This issue is fixed in version 1.22.0.
Weakness: CWE-125
Which versions are affected by CVE-2026-115920?
2 open source packages ship an affected version, including onnx.
| Package | Affected | Fixed in |
|---|---|---|
onnx PyPI | From 1.2.1 | 1.22.0 |
onnx PyPI | From 1.2.1 | 1.22.0 |
How do I fix CVE-2026-115920?
Upgrade onnx to 1.22.0 or later.
Is CVE-2026-115920 being exploited?
There is no known exploitation of CVE-2026-115920 in the wild and no public exploit code.
Sources
Also known as DEBIAN-CVE-2026-115920, ECHO-36c8-f2f9-8140, GHSA-6pwm-x34p-fvj2, UBUNTU-CVE-2026-115920.