Log in Start free

Vulnerability · Published Oct 26, 2026 · Assigned by GitHub_M

CVE-2026-115920ONNX: Out-of-bounds read in ConvTranspose shape inference via undersized weight rank

Medium 6.2

In plain English

A medium-severity out-of-bounds read in onnx from onnx. There are no confirmed reports of exploitation so far. The chance it is exploited in the next 30 days is estimated at 0.2% (higher than 7% of all known vulnerabilities). Fixed versions are available - upgrade the affected package.

Severity (CVSS 3.1)6.2Medium
Exploit odds (30 days)0.2%top 92.5% of all CVEs
Exploited in the wildNo reportsCISA SSVC: none
Attack toolingNoneno public tooling known

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.

Affected packages & fixed versions

EcosystemPackageFixed inAdvisory
Debian:12onnxno fix listedDEBIAN-CVE-2026-115920
Debian:13onnxno fix listedDEBIAN-CVE-2026-115920
Debian:14onnxno fix listedDEBIAN-CVE-2026-115920
PyPIonnx1.22.0GHSA-6pwm-x34p-fvj2
Ubuntu:22.04:LTSonnxno fix listedUBUNTU-CVE-2026-115920
Ubuntu:26.04:LTSonnxno fix listedUBUNTU-CVE-2026-115920
Ubuntu:26.10onnxno fix listedUBUNTU-CVE-2026-115920

Affected products (as reported)

VendorProductVersions
onnx onnx >= 1.2.1, < 1.22.0

Weakness: CWE-125 · Vector: CVSS:3.1/AV:L/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H

References