Views
No views yet
.pt file achieves Arbitrary Code Execution via torch.load(),
while bypassing both picklescan 1.0.4 and modelscan 0.8.8.marshal.loads + types.FunctionType + importlib.import_module chain.
All three globals are Suspicious (not Dangerous) in picklescan,
and absent from modelscan unsafe_globals.1import torch
2torch.load('malicious_model.pt', weights_only=False) # ACE