Mlflow Unauthenticated Access Scanner

MLflow is an open-source platform designed to manage the complete machine learning lifecycle, from experimentation to production deployment and monitoring. It is widely used by data scientists and machine learning engineers in various industries to streamline their workflows. Companies leverage MLflow to track and organize their machine learning experiments, enabling reproducibility and collaborative research efforts. A well-structured MLflow deployment can significantly enhance productivity and facilitate model sharing among teams.

Unauthenticated Access vulnerability occurs when a system unintentionally allows users to interact with sensitive resources without verifying their identity. This flaw arises from misconfigurations in security settings, often leading to unauthorized exposure of critical data or functionality. In the context of MLflow, such access can result in unintentional information leakage, manipulation of experiments, and potential data breaches.

Specifically, this scanner targets the MLflow Tracking Server's REST API endpoints, such as /api/2.0/mlflow/experiments/search and /api/2.0/mlflow/runs/search, which may be exposed without authentication. Attackers can enumerate experiments, runs, and registered models, and even create or delete resources if the server allows write operations without credentials. The vulnerability often stems from default configurations that disable authentication or from reverse proxy misconfigurations.

If exploited, an attacker can gain full visibility into sensitive machine learning data, including model parameters, metrics, and artifacts. They could tamper with experiments, inject malicious models, or exfiltrate proprietary algorithms. This could lead to intellectual property theft, compromised model integrity, and regulatory non-compliance, especially in industries handling sensitive data like healthcare or finance.