Miggo at BlackHat 2026!
Easily map the attack path and prioritize which CVEs are a threat to your organization
Miggo AIGenerate vendor-ready rules for the observed attack patterns, plus reasoning and safe deployment guidance
| Package Name | Ecosystem | Vulnerable Versions | First Patched Version |
|---|---|---|---|
| mlflow | pip | <= 3.10.1 | 3.11.0rc0 |
Miggo AIRoot Cause AnalysisThe vulnerability is an authorization bypass in MLflow's server, allowing users to access model artifacts without proper permissions. The analysis of the patch commit reveals that the fix involves adding authorization checks for two specific functionalities: downloading a model artifact via an AJAX endpoint and listing model artifacts via an API endpoint. The patch was applied in mlflow/server/auth/__init__.py, which acts as a central point for authorization logic. It adds validation rules to dictionaries that map request paths or types to validation functions. The absence of these rules for the artifact-related endpoints was the root cause of the vulnerability. The vulnerable functions are the handlers in mlflow/server/handlers.py that were responsible for processing these requests (_get_logged_model_artifact_handler and _list_logged_model_artifacts_handler). Before the patch, these handlers would be executed without any prior authorization check, leading to the information disclosure.
