Live › vulnerability › GHSA-64q6-5qv7-cwj9
Go to dashboardGluonTS, an open source Python library for deep learning time series models, executes arbitrary operating system commands when a model directory is deserialized through Predictor.deserialize() or RepresentablePredictor.deserialize(). An attacker who supplies a crafted serialized model directory gets commands run with the privileges of the process that loads it.
Summary written from the sources below. Check the originals before acting.
First seen by InfraStack 2026-10-24 18:56 UTC. All times on this page are UTC.
Teams that load GluonTS predictors with deserialize() from a model directory they do not fully control, such as user-supplied models, shared model stores, or artifacts downloaded from outside sources. Systems that only load models they trained and stored themselves are much less exposed, though the sources do not say how likely a tampered artifact is.
No product ranges from the sources yet.
pip show gluonts and treat anything below 0.17.0 as affected.Predictor.deserialize( or .deserialize( on GluonTS predictors, especially where the model directory comes from users, uploads, or external storage.0.17.0 or later with pip install --upgrade gluonts>=0.17.0.Each row is the first time that source carried this signal. The analyst’s rows are its own findings.
CVE-2026-100308 - GluonTS arbitrary command execution during model deserialization
Bulletin ID: 2026-119-AWS Scope: AWS Content Type: Important (requires attention) Publication Date: 09/29/2026 08:00 AM PDT
Description:
GluonTS is an open source library for deep learning based time series models. We identified CVE-2026-100308 that allows arbitrary command execution upon deserialization of untrusted model artifacts.
Deserialization of untrusted data in the model loading component in Amazon GluonTS before 0.17.0 might allow context-dependent attackers to execute arbitrary operating system commands with the privileges of the loading process via a crafted serialized model directory. The issue arises whenever a user calls Predictor.deserialize() or any RepresentablePredictor.deserialize() on a model directory they do not fully control.
Impacted versions: <0.17.0
Resolution:
This issue has been addressed in GluonTS version 0.17.0. We recommend upgrading to the latest version and ensuring any forked or derivative code is patched to incorporate the new fixes.
Workarounds:
We recommend users to only deserialize trusted model artifacts using the library.
References:
Acknowledgement:
We would like to thank Michael Holmquist (Hasp Labs) for collaborating on this issue through the coordinated vulnerability disclosure process.
Please email [email protected] with any security questions or concerns..
This page as .md · .json · in your agent: get_signal("GHSA-64q6-5qv7-cwj9")