CVE-2024-25723

8.8 HIGH

📋 TL;DR

This vulnerability in ZenML Server allows remote attackers to escalate privileges by activating user accounts with only a valid username and new password via the /api/v1/users/{user_name_or_id}/activate endpoint. It affects all ZenML deployments running vulnerable versions, particularly those exposed to untrusted networks. The flaw enables unauthorized account activation and potential administrative access.

💻 Affected Systems

Products:
  • ZenML Server
Versions: All versions before 0.42.2, 0.43.1, 0.44.4, and 0.46.7
Operating Systems: All
Default Config Vulnerable: ⚠️ Yes
Notes: Affects Python package installations of ZenML with the server component enabled.

📦 What is this software?

⚠️ Risk & Real-World Impact

🔴

Worst Case

Remote attacker gains administrative control over the ZenML Server, potentially compromising the entire ML pipeline, stealing sensitive data, and deploying malicious models.

🟠

Likely Case

Unauthorized user activation leading to privilege escalation, data exfiltration, and unauthorized access to ML workflows and artifacts.

🟢

If Mitigated

Limited impact with proper network segmentation and authentication controls, though the vulnerability still exists in the codebase.

🌐 Internet-Facing: HIGH
🏢 Internal Only: MEDIUM

🎯 Exploit Status

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

Exploitation requires knowledge of a valid username but no authentication, making it relatively straightforward for attackers with some reconnaissance.

🛠️ Fix & Mitigation

✅ Official Fix

Patch Version: 0.42.2, 0.43.1, 0.44.4, or 0.46.7

Vendor Advisory: https://www.zenml.io/blog/critical-security-update-for-zenml-users

Restart Required: Yes

Instructions:

1. Check current ZenML version with 'zenml version'. 2. Upgrade to patched version: 'pip install --upgrade zenml==0.46.7' (or appropriate version). 3. Restart ZenML Server services.

🔧 Temporary Workarounds

Disable vulnerable endpoint

all

Temporarily block access to the /api/v1/users/{user_name_or_id}/activate endpoint via web server configuration or firewall rules.

Example for nginx: location ~ ^/api/v1/users/.*/activate$ { deny all; }

Network isolation

linux

Restrict ZenML Server access to trusted IP addresses only using firewall rules.

Example for iptables: iptables -A INPUT -p tcp --dport 8080 -s TRUSTED_IP -j ACCEPT
iptables -A INPUT -p tcp --dport 8080 -j DROP

🧯 If You Can't Patch

  • Implement strict network access controls to limit ZenML Server exposure
  • Enable comprehensive logging and monitoring for suspicious activation attempts

🔍 How to Verify

Check if Vulnerable:

Check ZenML version: 'zenml version' or 'pip show zenml'. If version is before 0.42.2, 0.43.1, 0.44.4, or 0.46.7, system is vulnerable.

Check Version:

zenml version

Verify Fix Applied:

After upgrade, verify version is 0.42.2, 0.43.1, 0.44.4, or 0.46.7+ with 'zenml version'. Test that user activation requires proper authentication.

📡 Detection & Monitoring

Log Indicators:

  • Unusual POST requests to /api/v1/users/*/activate endpoint
  • User activation events from unexpected IP addresses
  • Failed authentication attempts followed by activation requests

Network Indicators:

  • HTTP POST requests to user activation endpoints without proper authentication headers
  • Traffic patterns suggesting privilege escalation attempts

SIEM Query:

source="zenml-server" AND (url_path="/api/v1/users/*/activate" OR message="user activation")

🔗 References

📤 Share & Export