CVE-2024-27320
📋 TL;DR
This vulnerability allows remote code execution in Refuel Autolabel library when processing malicious CSV files. Attackers can execute arbitrary Python code by crafting CSV files that get passed to eval() during classification tasks. Users of Autolabel versions 0.0.8 and newer who process untrusted CSV files are affected.
💻 Affected Systems
- Refuel Autolabel
📦 What is this software?
Autolabel by Refuel
⚠️ Risk & Real-World Impact
Worst Case
Complete system compromise allowing attacker to execute arbitrary commands, install malware, exfiltrate data, or pivot to other systems.
Likely Case
Data theft, system manipulation, or deployment of ransomware/cryptominers on affected systems.
If Mitigated
Limited impact if only trusted CSV files are processed and proper input validation is implemented.
🎯 Exploit Status
Exploitation requires victim to process a malicious CSV file. Public advisory includes technical details making exploitation straightforward.
🛠️ Fix & Mitigation
✅ Official Fix
Patch Version: Check vendor advisory for latest patched version
Vendor Advisory: https://hiddenlayer.com/sai-security-advisory/2024-09-autolabel/
Restart Required: No
Instructions:
1. Check current Autolabel version. 2. Update to latest patched version via pip: pip install --upgrade autolabel. 3. Verify no vulnerable code remains in deployment.
🔧 Temporary Workarounds
Disable CSV file processing
allTemporarily disable classification tasks that accept CSV files until patched.
Implement CSV validation
allAdd strict validation to reject CSV files containing Python code patterns.
🧯 If You Can't Patch
- Restrict CSV file uploads to trusted sources only
- Implement network segmentation to isolate Autolabel instances from sensitive systems
🔍 How to Verify
Check if Vulnerable:
Check Autolabel version: python -c "import autolabel; print(autolabel.__version__)". If version >= 0.0.8, system is vulnerable.Check Version:
python -c "import autolabel; print(autolabel.__version__)"Verify Fix Applied:
Verify version is updated beyond vulnerable range and test with safe CSV files to ensure functionality.📡 Detection & Monitoring
Log Indicators:
- Unusual Python execution errors
- CSV processing failures with eval-related errors
- Unexpected system commands in logs
Network Indicators:
- Unexpected outbound connections from Autolabel instances
- Data exfiltration patterns
SIEM Query:
source="autolabel.log" AND ("eval" OR "exec" OR "code injection")