AIKIDO-2026-188932
modelscope is vulnerable to Deserialization of Untrusted Data
87
High Risk
This Affects:
modelscopeTL;DR
ModelScope parses model and dataset configuration files with PyYAML's unsafe yaml.Loader, which constructs arbitrary Python objects from YAML tags. When a model repository is loaded, a poisoned configuration file containing Python object construction tags is deserialized during model resolution. This lets externally controlled repository content run arbitrary code on the machine loading the model. The remediation parses configuration YAML with a safe loader that does not instantiate arbitrary Python objects.
Who does this affect?
You are affected if you are using a version that falls within the vulnerable range and you load MPlug, HiTeA, or Sambert TTS models (or otherwise parse their repository config.yaml with ModelScope) from untrusted or externally controlled repositories.
Background info
modelscope is vulnerable to Deserialization of Untrusted Data in versions 1.0.4 - 1.40.0.
How to fix this
Upgrade the modelscope library to the patch version.
Links
CVE Detail
Related Issue
Other
github.com/modelscope/modelscopegithub.com/modelscope/modelscope/blob/v1.40.0/modelscope/models/audio/tts/voice.pygithub.com/modelscope/modelscope/blob/v1.40.0/modelscope/models/multi_modal/mplug/configuration_mplug.py
vulncheck.com/advisories/modelscope-through-1.40.0-unsafe-yaml-deserialization-in-model-config-loading
Are You Affected?
Connect your repositories to instantly see whether vulnerable or malicious packages exist in your codebase.
Free. No credit card required.
I consent to receiving marketing communications based on Aikido’s Privacy Policy.
SOC 2Compliant
ISO 27001CompliantISO 42001Compliant
FedRAMPImplementing© All Intel data is openly available and commercially licensed.