CVE-2026-55832
Received Received - Intake

Path Traversal in Tract ONNX Inference Toolkit

Publication date: 2026-09-14

Last updated on: 2026-09-14

Assigner: GitHub, Inc.

Description
Tract is a tiny, no-nonsense, self-contained TensorFlow and ONNX inference toolkit. Prior to 0.21.17, 0.22.3, and 0.23.2, the tract-onnx crate passes the attacker-controlled external_data location from an ONNX model through onnx/src/tensor.rs get_external_resources and joins the value to the model directory without rejecting absolute paths or parent directory components. Loading an untrusted model through model_for_path can therefore make onnx/src/data_resolver.rs MmapDataResolver open an arbitrary local file and place the file contents into model tensors or inference output. Attacker-controlled offset and length fields can also select an out-of-range mapping slice and cause a denial of service, but the flaw does not write files or execute code. This issue is fixed in versions 0.21.17, 0.22.3, and 0.23.2.

CVSS Scores

EPSS Scores

Probability:
Percentile:

Meta Information

Published
2026-09-14
Last Modified
2026-09-14
Generated
2026-09-14
AI Q&A
2026-09-14
EPSS Evaluated
N/A
NVD
CVE-2026-55832
EUVD
EUVD-2026-77721

Affected Vendors & Products

Showing 4 associated CPEs
Vendor Product Version / Range
tract tract to 0.21.17 (inc)
tract tract to 0.22.3 (inc)
tract tract to 0.23.2 (inc)
sonos tract to 0.23.2 (exc)

Helpful Resources

Exploitability

CWE
CWE Icon
KEV
KEV Icon
CWE ID Description
CWE-22 The product uses external input to construct a pathname that is intended to identify a file or directory that is located underneath a restricted parent directory, but the product does not properly neutralize special elements within the pathname that can cause the pathname to resolve to a location that is outside of the restricted directory.

Attack-Flow Graph

AI Quick Actions

Instant insights powered by AI
Executive Summary

CVE-2026-55832 is a path traversal vulnerability in the Tract ONNX inference toolkit. It occurs when loading untrusted ONNX models. The toolkit improperly handles the external_data location field by joining it with the model directory without validation. This allows crafted models to read arbitrary local files using absolute paths or directory traversal sequences like ...

The flaw exists in versions before 0.21.17, 0.22.3, and 0.23.2. Attackers can exploit this by embedding malicious paths in ONNX files, causing the toolkit to access unintended files. The vulnerability does not allow file writes or code execution but can expose sensitive data.

Detection Guidance

Detecting this vulnerability requires checking if your system uses vulnerable versions of the Tract library (prior to 0.21.17, 0.22.3, or 0.23.2). Inspect installed packages with commands like 'cargo tree -p tract-onnx' or 'pip show tract-onnx'. If vulnerable, look for suspicious ONNX model files that may contain crafted external_data paths.

Impact Analysis

This vulnerability could allow an attacker to read sensitive files on your system if you process untrusted ONNX models. The files could include system files like /etc/passwd or application data containing secrets. It may also cause crashes by reading out-of-bounds memory regions.

The impact depends on what files are accessible to the process using Tract. If the process has high privileges, the attacker could access system-wide sensitive files. Even with limited privileges, user data or configuration files might be exposed.

Compliance Impact

This vulnerability could lead to violations of GDPR, HIPAA, and other privacy regulations. Unauthorized access to personal or health data through file reads would constitute a data breach. Organizations must report such breaches under these regulations, potentially resulting in fines or penalties.

Compliance programs relying on Tract for ONNX model processing may need to update to patched versions immediately. Failure to address this could demonstrate negligence in protecting sensitive data, increasing legal and regulatory risks.

Mitigation Strategies

Upgrade the Tract library to versions 0.21.17, 0.22.3, or 0.23.2 or later. Avoid loading untrusted ONNX models until patched. If immediate upgrade isn't possible, restrict model loading to trusted sources only.

Chat Assistant

Ask questions about this CVE
Hi! I’m here to help you understand CVE-2026-55832. Ask me anything about the vulnerability, its impact, or mitigation strategies.
0/70

EPSS Chart