---
id: "CVE-2026-0596"
page: "https://infrastack.ai/live/CVE-2026-0596"
kind: "vulnerability"
aliases: ["CVE-2026-0596", "GHSA-rvhj-8chj-8v3c"]
title: "Mlflow: Command Injection when serving models with enable_mlserver=True"
severity: "critical"
cvss: 9.6
epss: 0.01393
exploited: false
status: "open"
category: "ai-tooling"
products: [{"name": "mlflow", "fixed": ["3.9.0"], "affected": "< 3.9.0", "ecosystem": "pypi"}, {"name": "lfprojects mlflow", "fixed": [], "affected": "", "ecosystem": "vendor"}, {"via": "cpe", "name": "mlflow", "fixed": [], "affected": "", "ecosystem": "pypi"}]
ecosystems: ["pypi", "vendor"]
tags: ["ai-tooling", "cwe-78", "ml-platform", "mlflow", "model-serving", "python"]
not_affected_if: ["fixed-version", "mlserver-disabled", "trusted-model-uri"]
published_at: "2026-03-31T15:16:10+00:00"
first_seen_at: "2026-11-16T15:37:43+00:00"
updated_at: "2026-11-16T16:08:50+00:00"
sources: [{"source": "ghsa", "url": "https://github.com/advisories/GHSA-rvhj-8chj-8v3c", "first_seen_at": "2026-11-16T15:37:43+00:00"}, {"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-0596", "first_seen_at": "2026-11-16T16:08:50+00:00"}]
exploit_sources: [{"type": "poc", "url": "https://huntr.com/bounties/2e905add-f9f5-4309-a3db-b17de5981285"}, {"type": "epss", "score": 0.01393, "url": "https://api.first.org/data/v1/epss?cve=CVE-2026-0596"}]
---

# Mlflow: Command Injection when serving models with enable_mlserver=True

Link for people: https://infrastack.ai/live/CVE-2026-0596 (this .md is for agents)

When MLflow serves a model with enable_mlserver=True, the model_uri is placed into a bash -c shell command without sanitization. A model URI containing shell substitutions such as $() or backticks lets an attacker run commands of their choosing.

_Summary, check and fix written by a model from the sources; labelled as such._

## Affected

- pypi:mlflow affected < 3.9.0 fixed 3.9.0
- vendor:lfprojects mlflow affected range not listed fixed no fixed version listed
- pypi:mlflow affected range not listed fixed no fixed version listed

## How to check

- Run `pip show mlflow` and compare the version against `3.9.0`.
- Search deployment code and scripts for `enable_mlserver=True` or `enable_mlserver`.
- Check whether model URIs come from users or untrusted sources.

## Fix

- `mlflow` `3.9.0`

## Timeline

- 2026-11-16T15:37:43+00:00 ghsa https://github.com/advisories/GHSA-rvhj-8chj-8v3c
- 2026-11-16T16:08:50+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-0596

## Documents

- [fix] Patch: github.com https://github.com/mlflow/mlflow/pull/19738
- [fix] Patch: github.com https://github.com/mlflow/mlflow/commit/202fac4c83ccc8544c087c142b80196d0e60695c
- [poc] Exploit: huntr.com https://huntr.com/bounties/2e905add-f9f5-4309-a3db-b17de5981285

## From the sources

A command injection vulnerability exists in mlflow/mlflow when serving a model with `enable_mlserver=True`. The `model_uri` is embedded directly into a shell command executed via `bash -c` without proper sanitization. If the `model_uri` contains shell metacharacters, such as `$()` or backticks, it allows for command substitution and execution of attacker-controlled commands. This vulnerability affects the latest version of mlflow/mlflow and can lead to privilege escalation if a higher-privileged service serves models from a directory writable by lower-privileged users.

---

_Free to read, with fair use of about 60 pages a minute. For higher limits or API access: sales@infrastack.ai_
