CVE-2026-115919
Awaiting Analysis Awaiting Analysis - Queue

Heap Buffer Overflow in ONNX Prior to 1.22.0

Publication date: 2026-10-26

Last updated on: 2026-10-26

Assigner: GitHub, Inc.

Description
Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. Prior to 1.22.0, RNNShapeInference_opset1_to_6() processes malformed RNN and LSTM nodes from opsets 1 through 6 and GRU nodes from opsets 3 through 6 without verifying that the first input has rank three. The function accesses dim(0) and dim(1) on an undersized protobuf shape, which can read outside the intended repeated field and terminate an application when onnx.shape_inference.infer_shapes() processes an untrusted model that can pass onnx.checker.check_model(). This issue is fixed in version 1.22.0.

CVSS Scores

EPSS Scores

Probability:
Percentile:

Meta Information

Published
2026-10-26
Last Modified
2026-10-26
Generated
2026-10-26
AI Q&A
2026-10-26
EPSS Evaluated
N/A
NVD
CVE-2026-115919
EUVD
EUVD-2026-110040

Affected Vendors & Products

Showing 1 associated CPE
Vendor Product Version / Range
onnx onnx < 1.22.0

Helpful Resources

Exploitability

CWE
CWE Icon
KEV
KEV Icon
CWE ID Description
CWE-125 The product reads data past the end, or before the beginning, of the intended buffer.

Attack-Flow Graph

AI Quick Actions

Instant insights powered by AI
Executive Summary

This is an out-of-bounds read vulnerability in ONNX's shape inference function for legacy RNN, GRU, and LSTM nodes (opsets 1-6). The RNNShapeInference_opset1_to_6() function accessed tensor dimensions without verifying the input had sufficient rank, causing crashes when processing malformed models.

Detection Guidance

To detect this vulnerability, check the ONNX version installed on your system. Run: onnx.__version__ in Python or check package managers like pip show onnx. If the version is below 1.22.0, the system is vulnerable. Additionally, inspect models using opsets 1-6 for malformed RNN, GRU, or LSTM nodes that may trigger the out-of-bounds read.

Impact Analysis

The vulnerability could cause application crashes (denial of service) when processing specially crafted untrusted models. It requires local access but no privileges or user interaction. Modern ONNX versions (1.22.0+) are not affected.

Mitigation Strategies

Upgrade ONNX to version 1.22.0 or later immediately. Avoid processing untrusted models with opsets 1-6 until upgraded. If upgrading is not possible, restrict model processing to trusted sources only.

Chat Assistant

Ask questions about this CVE
Hi! I’m here to help you understand CVE-2026-115919. Ask me anything about the vulnerability, its impact, or mitigation strategies.
0/70

EPSS Chart