MLflow CVE-2026-64849: Exploited Unauthenticated SSRF
If an MLflow tracking server below 3.15.0 is reachable by anyone you do not fully trust, assume attackers can read your cloud instance credentials through it: upgrade today, then check webhooks and cloud audit logs.
Must act: Teams running MLflow tracking servers 3.3.0 through 3.14.x bound beyond localhost, especially on cloud VMs or Kubernetes nodes with an attached IAM role, service account or managed identity.
Can relax: Local-only tracking servers on the default 127.0.0.1 binding, servers below 3.3.0 (no webhooks feature, though they need upgrading for other reasons), and anyone already on 3.16.0 or later.
- Check the version on every tracking server (pip show mlflow) and upgrade to 3.16.0 or later; 3.17.2 is current.
- Until upgraded, take the server off public networks or put it behind a VPN or identity-aware proxy.
- Block the MLflow host from reaching 169.254.169.254 and other link-local metadata addresses, and enforce IMDSv2 on AWS.