CVE-2026-69148 - MLflow Vulnerability Exposes User Artifacts
CVE-2026-69148 - MLflow Vulnerability 🚨
MLflow is an open-source AI engineering platform for agents, large language models, and machine learning models. Prior to version 3.15.0, the CreateModelVersion function accepts a run_id or model_id after _validate_source_run() or _validate_source_model() in mlflow/server/handlers.py. This verification only checks for path containment, allowing authenticated users to create a model version that references another user’s artifact directory and read files through GET /model-versions/get-artifact without the required READ permission. This issue has been fixed in version 3.15.0.
Vulnerability Details
The CVE-2026-69148 vulnerability affects MLflow versions prior to 3.15.0. The description provided for CVE-2026-69148 states: “MLflow is an open-source AI engineering platform for agents, large language models, and machine learning models. Prior to 3.15.0, CreateModelVersion accepts a run_id or model_id after _validate_source_run() or _validate_source_model() in mlflow/server/handlers.py verifies only path containment, allowing authenticated users to create a model version that references another user’s artifact directory and read files through GET /model-versions/get-artifact without the required READ permission.”
CVSS Score
The vulnerability has been assigned a CVSS V3.1 score of AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:L/A:N. This score indicates that:
- Confidentiality is rated as high (C:H)
- Integrity is rated as low (I:L)
- The attack vector is network (AV:N)
- Attack complexity is low (AC:L)
- Requires low privileges (PR:L)
- No user interaction is needed (UI:N)
- The scope is unchanged (S:U)
- Availability is none (A:N)
This vulnerability is categorized under CWE-862, which refers to missing authorization.
References
This new CVE was received by [email protected] on August 17, 2026. For more information, you can check the following references:
The fix for this issue is included in version 3.15.0.
For the complete article, see: Read full article