Aug 17, 2026 · ZeroDay Catalog
Critical Server-Side Request Forgery Vulnerability in MLflow
A server-side request forgery vulnerability has been identified in MLflow versions prior to 3.15.0, potentially exposing sensitive internal services.
What changed
A server-side request forgery (SSRF) vulnerability was discovered in MLflow, affecting versions from 3.3.0 up to 3.14.0. The vulnerability arises from the handling of webhook URLs, which allows unauthorized users to exploit the system and access internal or cloud metadata services.
The issue was addressed in version 3.15.0, which mitigates the risk by ensuring that the validated address is pinned and not subject to redirection.
Who is affected
Organizations utilizing MLflow for AI engineering and machine learning model management are at risk, particularly those running versions prior to 3.15.0.
The vulnerability could impact any deployment of MLflow that uses the affected webhook endpoint, potentially exposing sensitive internal services.
- MLflow users on versions 3.3.0 to 3.14.0
- Organizations leveraging MLflow for AI and machine learning projects
- Developers integrating MLflow with internal systems or cloud services
What to do
It is imperative for users of MLflow to upgrade to version 3.15.0 or later to mitigate the SSRF vulnerability. This update will ensure that the system is protected against unauthorized access to internal services.
Additionally, organizations should review their webhook configurations and monitor for any unusual access patterns that may indicate exploitation attempts.
- Upgrade MLflow to version 3.15.0 or later immediately.
- Review and secure webhook configurations to prevent unauthorized access.
- Monitor system logs for any suspicious activity related to webhook endpoints.