CVE-2024-25723

Vulnerability

Overview

The ZenML machine learning package for Python contains a privilege escalation vulnerability in its server component. The issue resides in the /api/v1/users/{user_name_or_id}/activate REST API endpoint, which allows an attacker to reset the password for any existing user by simply providing a valid username and a new password in the request body [1][2][4]. This design flaw bypasses any authentication or authorization checks, enabling unauthorized account takeover.

Exploitation and

Attack Surface

Exploitation requires only knowledge of a valid username within the ZenML deployment. An attacker can send a crafted HTTP POST request to the activate endpoint with the target username and a password of their choice. The endpoint processes the request without verifying the requester's identity, effectively allowing anyone to assume control of any user's account [1][4]. No prior authentication or session is needed, making the attack remotely exploitable.

Impact

Successful exploitation grants the attacker full control over the compromised ZenML account, including access to any ML pipelines, models, data, and infrastructure configurations managed by that user. This can lead to data exfiltration, model poisoning, or disruption of production ML workflows. The vulnerability is especially critical for organizations using ZenML in shared or multi-tenant environments.

Mitigation

Status

The vulnerability is confirmed to affect all ZenML versions prior to 0.46.7, with the exception of patched versions 0.44.4, 0.43.1, and 0.42.2 [1][3][4]. Users are strongly advised to upgrade to version 0.46.7 or one of the patched releases immediately. ZenML Cloud users do not require action as the platform has already been updated [4].