Advisory record · GHSA-hqmj-h5c6-369m

ONNX Untrusted Model Repository Warnings Suppressed by silent=True in onnx.hub.load() — Silent Supply-Chain Attack

CVSS 8.6 HighEPSS 0.0033 · 26th percentile

Published by the GitHub Advisory Database at 16 Mar 2026, 16:23 UTC. agentprime has not analysed this yet, so whether it affects any app is not known.

PackageEcosystemAffectedFixed in
onnxPyPI<1.21.0rc11.21.0rc1
Identifiers
GHSA-hqmj-h5c6-369m · CVE-2026-28500 · PYSEC-2026-103
Weakness (CWE)
CWE-345 · CWE-494 · CWE-693 · CWE-829
Published
16 Mar 2026
Updated
10 Sept 2026

Machine-written summary

Summary written by Claude Haiku from the source advisory. Not reviewed by a person.

The ONNX package (versions before 1.21.0rc1) lets a caller pass silent=True to onnx.hub.load(), which suppresses the trust warnings and user prompts that normally appear when a model comes from an unverified GitHub repository. An attacker who controls such a repo can have a model loaded without the user being aware, and if that model runs code on load, the attacker can run code on the machine. The source notes the SHA256 check is ineffective here because its manifest comes from the same attacker-controlled repo. Upgrade to 1.21.0rc1 or later, where the feature was removed.

Written 5 Nov 2026.

Not analysed yet

We haven’t published our analysis of this Advisory yet. Once a person has reviewed it, this page will say what’s vulnerable, what has to be true for your app to be affected, and how to check.

Severity and exploitation

ScoreRatingCVSSVectorSource
8.6 High 3.1 CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:N/A:N CNA (GitHub_M)
  • CISA KEV Not listed.
  • EPSS 0.0033, the 26th percentile, on 5 Nov 2026: FIRST’s estimate of the chance of exploitation in the next 30 days.

Timeline

  1. 31 Oct 2026, 14:34 UTC First seen by agentprime
  2. 31 Oct 2026, 14:34 UTC Confirmed by OSV

Sources

Facts on this page come from the GitHub Advisory Database (CC-BY 4.0), OSV and the other sources listed, as published. We don’t reproduce full advisory descriptions, which can contain exploit code.