CVE-2026-116690: high-severity vulnerability in mlflow
MLflow: Arbitrary file read on unauthenticated MLflow servers via prompt model version source
Published · Updated
43Vexday Risk Score
Prioritize patching. It exploitation observed by VulnCheck.
ssvc Actcvss 7.5epss 0.6%
from disclosure to weapon
Published on NVDOct 27
VulnCheck+1d
exploitation probability
0.6%top 54% of all CVEs
observed exploitation
yesVulnCheck
MLflow is an open source AI engineering platform for agents, large language models, and machine learning models. From 3.10.0 until 3.17.0, the prompt branch in _create_model_version() accepts schemeless relative sources because path validation runs only when a source has a URL scheme, and mlflow.prompt.is_prompt is interpreted by tag presence rather than a true value. A remote caller can store a server-local path as the model version artifact location and use /api/2.0/mlflow/model-versions/get-artifact to read files accessible to the MLflow server process. Default unauthenticated servers require no privileges, while basic-auth deployments are exposed to authenticated users who can create registered models. This issue is fixed in version 3.17.0.
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N
Affected products
mlflow · mlflowRelated CVEs — mlflow
In the same product, most dangerous first.
CVE-2026-64849CRITICALMLflow: Unauthenticated full-read SSRF in webhook delivery: _validate_webhook_url bypassed via unvalidated HTTP redirects (and DNS rebinding)EPSS 10.0%KEVCVE-2026-116692CRITICALMLflow: Remote code execution on unauthenticated MLflow servers via third-party scorer deserialization importing caller-placed artifact modulesEPSS 0.7%CVE-2026-53995HIGHMLflow 3.17.1 and prior Path Traversal via Trace ID APIEPSS 0.6%CVE-2026-79721HIGHCVE-2026-79721EPSS 0.5%CVE-2026-69146MEDIUMMLflow: LogInputs endpoint bypasses per-run UPDATE authorization in basic-authEPSS 0.4%CVE-2026-69148HIGHMLflow: CreateModelVersion source validation does not check READ permission on referenced run_idEPSS 0.4%