DEV Community

CVE Reports
CVE Reports

Posted on • Originally published at cvereports.com

CVE-2026-69148: CVE-2026-69148: Broken Object Level Authorization (BOLA) in MLflow Model Registry

CVE-2026-69148: Broken Object Level Authorization (BOLA) in MLflow Model Registry

Vulnerability ID: CVE-2026-69148
CVSS Score: 7.1
Published: 2026-08-17

MLflow prior to version 3.15.0 fails to perform proper authorization checks when registering model versions, allowing authenticated users with access to a registered model to link and access artifacts from runs and models belonging to other users without authorization.

TL;DR

A Broken Object Level Authorization (BOLA) vulnerability in MLflow's Model Registry allows authenticated users to read private artifacts of arbitrary runs by registering them under their own model namespace.


Technical Details

  • CWE ID: CWE-862
  • Attack Vector: Network
  • CVSS: 7.1 (High)
  • EPSS Score: Not Registered
  • Impact: Confidentiality (High), Integrity (Low)
  • Exploit Status: None
  • KEV Status: Not Listed

Affected Systems

  • MLflow Server
  • mlflow: < 3.15.0 (Fixed in: 3.15.0)

Code Analysis

Commit: 4bb7474

Fix BOLA vulnerability on model version creation by validating read permission on source run/model

Mitigation Strategies

  • Upgrade MLflow server and client implementations to version 3.15.0 or later.
  • Enforce explicit backend cloud storage access boundaries via IAM policies to restrict broad file-system visual mappings.
  • Utilize reverse proxies or API gateways to inspect and drop unauthorized references to non-workspace entities.

Remediation Steps:

  1. Determine currently running MLflow versions in active development and staging configurations.
  2. Update standard dependencies to require mlflow>=3.15.0.
  3. Redeploy server container instances containing the patched library code.
  4. Verify authorization policies are active and execute an administrative audit of currently registered models and mapped sources.

References


Read the full report for CVE-2026-69148 on our website for more details including interactive diagrams and full exploit analysis.

Top comments (0)