Repository for the scripts presented in the blogpost: https://splint.gitbook.io/cyberblog/security-research/tersorflow-remote-code-execution-with-malicious-model
exploit.py
: used to create a simple malicious model with a reverse shellinject.py
: injects the malicious layer in a legitimate modelmodel.py
: simulate the usage ofload_mode
on a malicious modeldetector.py
: detects a maliciousLambda
in a.h5
model
The contents of this repository are exclusively for research and entertainment purposes.