CVE-2026-129266 · AI agents & MCP
NO FIX YET · SEE WHAT TO DO
Deep Java Library (DJL), an open-source Java deep learning framework, does not validate the resource location in model archive metadata. A remote unauthenticated attacker can use a crafted model archive to read any file the library process can access.
first public Nov 16, 2026 · 2 sources
Check it with your coding agent
It reads this page and tells you if your code or app is affected.
Applications that load model archives through DJL, especially djl-serving or the djl-inference container, which exposes model registration and inference APIs. Exposure is greatest where untrusted callers can register models or supply model URLs. Teams that only load models from sources they control and do not expose the APIs are less exposed, but the fix still applies.
Not affected if apis-not-network-exposed, only-trusted-model-sources, versions-0.39.0-or-later
Who cares aws, deep-learning, djl, djl-serving, java, lab-disclosure, model-serving
>=0.7.0 and <=0.38.0.model_url_pattern is set on djl-serving.0.39.0 or later.CVE-2026-129266 - Arbitrary Local File Read in DJL - AWS
Bulletin ID: 2026-183-AWS Scope: AWS/Amazon Content Type: Important (requires attention)/Informational Publication Date: 11/16/2026 10:30 AM PST
Description:
Deep Java Library (DJL) is an open-source, engine-agnostic Java framework for deep learning. We identified CVE-2026-129266, an issue where DJL doesn't validate the resource location specified in model archive metadata. A remote unauthenticated third party could use a crafted model archive to read arbitrary files that the library process can access.
Impacted versions: >=0.7.0, <=0.38.0
Resolution:
This issue has been addressed in DJL version 0.39.0. We recommend upgrading to the latest version and ensuring any forked or derivative code is patched to incorporate the new fixes.
Workarounds:
Only load model archives from sources you trust. If you run djl-serving, or a container image that bundles it such as the djl-inference Deep Learning Container, restrict network access to the management and inference APIs so that untrusted callers cannot register models, and set model_url_pattern to an allowlist of model locations you control to restrict models loaded through inference requests.
References:
Acknowledgement:
We would like to thank Oguzhan Akkaya for collaborating on this issue through the coordinated vulnerability disclosure process.
Please email aws-security@amazon.com with any s ecurity questions or concerns.
Are you affected?
Ask from here. InfraStack checks the affected versions against your stack and answers in your inbox.
InfraStack checks the affected versions against your stack and answers in Slack, with the evidence and the fix.
InfraStack checks it against your stack and answers on your report, with the evidence and the fix.
This page as .md · .json · CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X
NVD data is provided by NIST and is public domain. Posts quoted belong to their authors.