CVE-2025-10155

7.8 HIGH

📋 TL;DR

This vulnerability allows attackers to bypass security checks in picklescan by disguising malicious pickle files with PyTorch-related extensions. When these files are incorrectly marked as safe and loaded, they can execute arbitrary code. Users of picklescan versions up to 0.0.30 are affected.

💻 Affected Systems

Products:
  • mmaitre314 picklescan
Versions: up to and including 0.0.30
Operating Systems: All
Default Config Vulnerable: ⚠️ Yes
Notes: Vulnerability exists in default scanning logic when processing pickle files with PyTorch extensions.

📦 What is this software?

⚠️ Risk & Real-World Impact

🔴

Worst Case

Remote code execution leading to complete system compromise, data theft, or ransomware deployment.

🟠

Likely Case

Malicious code execution in the context of the picklescan process, potentially allowing lateral movement or persistence.

🟢

If Mitigated

Limited impact if proper input validation and file scanning controls are implemented.

🌐 Internet-Facing: MEDIUM - Requires user to process untrusted pickle files, but common in ML/AI workflows.
🏢 Internal Only: MEDIUM - Internal users processing untrusted pickle files remain vulnerable.

🎯 Exploit Status

Public PoC: ✅ No
Weaponized: UNKNOWN
Unauthenticated Exploit: ⚠️ Yes
Complexity: LOW

Exploitation requires tricking users into scanning malicious pickle files with specific extensions.

🛠️ Fix & Mitigation

✅ Official Fix

Patch Version: 0.0.31 or later

Vendor Advisory: https://github.com/mmaitre314/picklescan/security/advisories/GHSA-jgw4-cr84-mqxg

Restart Required: No

Instructions:

1. Update picklescan using pip: pip install --upgrade picklescan
2. Verify version is 0.0.31 or higher

🔧 Temporary Workarounds

Manual file extension validation

all

Implement custom validation to reject pickle files with PyTorch extensions before scanning.

# Python example: if filename.endswith(('.pt', '.pth')): reject_file()

🧯 If You Can't Patch

  • Restrict picklescan usage to trusted environments only
  • Implement network segmentation to limit potential lateral movement

🔍 How to Verify

Check if Vulnerable:

Check picklescan version: python -c "import picklescan; print(picklescan.__version__)"

Check Version:

python -c "import picklescan; print(picklescan.__version__)"

Verify Fix Applied:

Confirm version is 0.0.31 or higher using version check command

📡 Detection & Monitoring

Log Indicators:

  • Unexpected process execution from picklescan
  • Pickle file scanning errors with PyTorch extensions

Network Indicators:

  • Unusual outbound connections from picklescan process

SIEM Query:

process_name='picklescan' AND (file_extension='.pt' OR file_extension='.pth')

🔗 References

📤 Share & Export