TLDR InfoSec · LAUNCHES & TOOLS
MODELSCAN
Summary from the TLDR InfoSec newsletter
ModelScan is an open source tool by Protect AI that scans machine learning models for unsafe code. It supports various model formats like H5, Pickle, and SavedModel. By scanning models, you can protect against Model Serialization Attacks that could compromise your data and credentials.