JFrog Security Research
Model Threats
Discover
Follow JFrog Security
Home
Model Threats
Discover
Follow JFrog Security
Machine Learning Model Threat Categories
ZIPSLIP
Zip-based model attempting ZipSlip path traversal
TFLOW-MALOPS
TensorFlow SavedModel with malicious operators
TFLOW-LAMBDA
TensorFlow SavedModel with Lambda Layers containing malicious code
PYTORCH-MALCODE
PyTorch model with embedded malicious code
PYTORCH-GETATTR
PyTorch model using getattr maliciously
PICKLE-MALCODE
Pickle-based model with embedded malicious code
PICKLE-GETATTR
Pickle-based model using getattr maliciously
ONNX-BACKDOOR
ONNX model with architectural backdoor causing unexpected outputs
NOAUTOLOAD-SUSCODE
Model does not support code execution on load but contains suspicious code strings
KERAS-LAMBDA
Keras model with Lambda Layers containing malicious code
KERAS-CUSTOM
Keras model with Custom Layers calling malicious functions
H5-LAMBDA
TensorFlow H5 model with Lambda Layers containing malicious code
GGUF-SSTI
GGUF model attempting template injection for arbitrary code execution