CVE-2024-1560: mlflow vulnerable to Path Traversal
MLflow 2.9.2 and earlier let an authenticated user delete arbitrary directories on the server through the artifact deletion path. A double URL-decode defeats path validation. The advisory lists no patched version.
Published Apr 16, 2024 · Analysis Nov 2, 2026
What an attacker can do
The deletion route _delete_artifact_mlflow_artifacts and the function local_file_uri_to_path validate the user's path, but delete_artifacts in local_artifact_repo.py runs an extra unquote on it. An encoded traversal sequence therefore passes validation as harmless text and becomes a real traversal after the second decode. The delete then lands on a directory outside the artifact root.
This was an attempt to fix CVE-2023-6831 that left a gap. The CVSS vector (8.1) shows network access, low privileges, and high integrity and availability impact. The caller needs a valid account on the server. They can then remove directories the MLflow process can write, which may include other teams' artifacts, the backend store or the application itself. There is no confidentiality loss in the vector.
Shared tracking servers with many users are the main target.
Who is affected
The pip package mlflow at 2.9.2 and earlier. Servers with the artifact proxy routes (mlflow-artifacts) enabled and several users are most exposed. Check:
pip show mlflowmlflow
- Ecosystem
- PyPI
- Affected
- <= 2.9.2
- Fixed in
- No patched version yet
- Check
pip show mlflow
How to fix it
The advisory lists no patched version for this CVE, and gives no workaround. Check the project's release notes for a newer release before you rely on one; the sources here do not confirm which release closed this issue.
Until you have confirmed a fix, limit accounts on the tracking server, run the MLflow process as a user that can write only to its artifact directory, and keep backups of the artifact store and the backend database.
How to tell if you were hit
Look in server logs for DELETE requests to the mlflow-artifacts routes that carry percent-encoded characters such as %2e or %2f, or doubly encoded forms like %252e. Check for directories missing from the artifact root or from the host that the MLflow process could write. Restore from backup and review which account sent the request.
Original advisory
From the GitHub Advisory Database (CC BY 4.0).
A path traversal vulnerability exists in the mlflow/mlflow repository, specifically within the artifact deletion functionality. Attackers can bypass path validation by exploiting the double decoding process in the _delete_artifact_mlflow_artifacts handler and local_file_uri_to_path function, allowing for the deletion of arbitrary directories on the server's filesystem. This vulnerability is due to an extra unquote operation in the delete_artifacts function of local_artifact_repo.py, which fails to properly sanitize user-supplied paths. The issue is present up to version 2.9.2, despite attempts to fix a similar issue in CVE-2023-6831.
Questions
- What is the double decode in CVE-2024-1560?
The path is decoded once during validation, and delete_artifacts decodes it again. Encoded traversal text passes the check and turns into a traversal at the second decode.
- Does it need an account?
Yes. The CVSS vector lists low privileges, so the caller needs a valid login.
- Is there a fixed version?
The advisory lists none. It describes the bug as present up to 2.9.2 and as a bypass of the CVE-2023-6831 fix.
References
How this page is made: the data panel comes from the GitHub Advisory Database, CISA’s KEV catalog, the NVD and FIRST EPSS, and updates twice a day. The analysis is written by an AI model from the sources listed above and checked by automated rules before it is published. Check versions against the vendor's advisory before you act. Methodology
More MLflow vulnerabilities
All MLflow entriesCVE-2026-116690
MLflow vulnerable to arbitrary file read on unauthenticated servers via prompt model version source
high7.5Oct 27, 2026CVE-2026-116691
MLflow CrewAI autologging records LLM provider API keys in logged trace spans
medium6.5Oct 27, 2026CVE-2026-116692
MLflow vulnerable to remote code execution on unauthenticated servers via third-party scorer deserialization importing caller-placed artifact modules
critical9.8Oct 27, 2026CVE-2026-115886
MLflow: SSRF on unauthenticated MLflow servers via third-party scorer kwargs in the scorer invoke, register, and GenAI evaluate endpoints
medium6.5Oct 23, 2026CVE-2026-53995
MLflow FileStore trace IDs allow path traversal and arbitrary file writes
high8.2Oct 20, 2026GHSA-gqvg-gmmx-x4hm
MLFLOW_ALLOW_PICKLE_DESERIALIZATION=False safety control bypassed by mlflow.statsmodels flavor — RCE via crafted model artifact
high8.8Sep 1, 2026
