CVE-2026-34445 ONNX: Malicious ONNX models can crash servers by exploiting unprotected object settings.

What is CVE-2026-34445?

CVE-2026-34445 is a high-severity vulnerability in onnx, published on 1 Apr 2026. An attacker can exploit it over the internet without logging in, with no action from the victim.

Official description

Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. Prior to version 1.21.0, the ExternalDataInfo class in ONNX was using Python’s setattr() function to load metadata (like file paths or data lengths) directly from an ONNX model file. It didn’t check if the "keys" in the file were valid. Due to this, an attacker could craft a malicious model that overwrites internal object properties. This issue has been patched in version 1.21.0.

Weakness: CWE-915, CWE-400, CWE-20

Which versions are affected by CVE-2026-34445?

3 open source packages ship an affected version, including onnx. It affects 1 product.

Products include Linuxfoundation Onnx.

How do I fix CVE-2026-34445?

Upgrade onnx to 1.21.0 or later.

Is CVE-2026-34445 being exploited?

There is no known exploitation of CVE-2026-34445 in the wild and no public exploit code.

Other vulnerabilities in onnx

• CVE-2025-51480 Path traversal in n/a High 8.8
• CVE-2024-5187 Arbitrary File Overwrite in download_model_with_test_data in onnx/onnx High 8.8
• CVE-2026-27489 ONNX: Path Traversal via Symlink High 8.7
• CVE-2026-28500 ONNX Untrusted Model Repository Warnings Suppressed by silent=True in onnx.hub.load() — Silent Supply-Chain Attack High 8.6
• CVE-2022-25882 Directory Traversal in onnx High 7.5

Sources

• CVE record
• NVD
• OSV
• GitHub advisory
• How this page is made

Also known as AZL-81989, DEBIAN-CVE-2026-34445, ECHO-a26a-b5fe-3ab2, GHSA-538c-55jv-c5g9, PYSEC-2026-2240, UBUNTU-CVE-2026-34445.