// Security Advisory Triage

MLflow arbitrary file read on unauthenticated servers, patched in 3.17.0

GitHub advisories — pip · published 2026-10-27 21:41 UTC
ACTION NEEDED PATCH_NOW SECURITYAPI
WHAT CHANGEDGitHub published CVE-2026-116690 on 27 October 2026: MLflow servers without authentication allow arbitrary file read via prompt model version source. Severity rated high. First patched version is 3.17.0.
WHY IT MATTERSAn unauthenticated attacker can read arbitrary files from the MLflow server filesystem, exposing credentials, model artifacts, configuration, and any data the server process can access. MLflow is commonly deployed as a central gateway for model serving and experiment tracking, making this a direct path to production ML infrastructure.
WHAT TO DOUpgrade MLflow to 3.17.0 or later. If you cannot upgrade immediately, enable authentication on all MLflow servers to block unauthenticated access until patching.
Model-reported confidence 95% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-m65p-9j7q-6w33
+summary: MLflow vulnerable to arbitrary file read on unauthenticated servers via prompt model version source
+severity: high
+published_at: 2026-10-27T21:26:56Z
+cve_id: CVE-2026-116690
+vulnerabilities[].package.name: mlflow
+vulnerabilities[].first_patched_version: 3.17.0
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningMLflow is a widely-deployed ML model serving and tracking framework. Unauthenticated arbitrary file read on servers is exploitable and actionable.
signalsnew advisory published · high severity rating · arbitrary file read vulnerability · unauthenticated attack surface · MLflow is AI/ML gateway and model serving framework · patched version available (3.17.0) · affects dependency tree of ML infrastructure
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)