Software Path Traversal (CVE-2025-15036) - Patch Now
CVE-2025-15036
Attackers exploit a critical path traversal in MLflow to overwrite system files. Update to version 3.7.0 or later to block this sandbox escape and privilege escalation vector.
Patch now - CVE-2025-15036 is a critical path traversal in MLflow prior to 3.7.0 that lets an attacker overwrite arbitrary files and escape sandboxes via a malicious tar.gz archive. Upgrade immediately to block privilege escalation and host compromise.
Overview
A critical security vulnerability, identified as CVE-2025-15036, has been discovered in the MLflow open-source platform. This flaw is a path traversal vulnerability that could allow an attacker to overwrite critical files on the system or escape intended security boundaries.
Vulnerability Details
The vulnerability exists in the extract_archive_to_dir function within MLflow’s code, specifically in the file mlflow/pyfunc/dbconnect_artifact_cache.py. In versions prior to 3.7.0, this function does not properly validate the paths of files contained within a tar.gz archive during extraction.
In simple terms, when MLflow processes a specially crafted tar.gz file, it fails to check if the file paths inside the archive are trying to navigate outside the intended destination folder. This allows a malicious file path like ../../../etc/passwd to be accepted, leading to the extraction of the file to a completely different, unauthorized location on the server’s filesystem.
Impact and Risk
The impact of this vulnerability is severe (CVSS score: 9.6). An attacker who can supply a malicious archive file to a vulnerable MLflow instance could:
- Overwrite arbitrary files, potentially disrupting system operations or corrupting critical data.
- Escape the sandbox directory in multi-tenant or shared cluster environments, accessing or modifying files belonging to other users or the system itself.
- Achieve elevated privileges by overwriting system or application files, which could lead to a full compromise of the host.
This type of flaw is a common vector for serious breaches. For context on how such vulnerabilities can be exploited in real-world attacks, you can review historical incidents in our breach reports.
Remediation and Mitigation
The primary and most effective action is to update the MLflow installation immediately.
- Immediate Patching: Upgrade MLflow to version 3.7.0 or later. This version contains the necessary validation to prevent path traversal during archive extraction.
- Workaround (If Patching is Delayed): If an immediate upgrade is not possible, restrict the processing of tar.gz archive files from untrusted sources within your MLflow workflows. This is a temporary measure and does not eliminate the risk.
- General Security Hygiene: Always practice the principle of least privilege for service accounts running MLflow and maintain regular software updates. Staying informed on emerging threats is crucial; follow the latest developments in our security news section.
Organizations using MLflow, particularly in shared environments, should treat this as a high-priority update to prevent potential system compromise.
Never miss a critical vulnerability
Get real-time security alerts delivered to your preferred platform.
Related Advisories
A command injection vulnerability exists in MLflow's model serving container initialization code, specifically in the `_install_model_dependencies_to_env()` function. When deploying a model with `env_...
A vulnerability in MLflow's pyfunc extraction process allows for arbitrary file writes due to improper handling of tar archive entries. Specifically, the use of `tarfile.extractall` without path valid...
ColdFusion versions 2025.9, 2023.20 and earlier are affected by an Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal') vulnerability that could lead to arbitrary code execut...
Xerte Online Toolkits versions 3.15 and earlier contain an incomplete input validation vulnerability in the elFinder connector endpoint that fails to block PHP-executable extensions .php4 due to an in...
Other Lfprojects Mlflow Vulnerabilities
A command injection vulnerability exists in MLflow's model serving container initialization code, specifically in the `_install_model_dependencies_to_env()` function. When deploying a model with `env_...
A vulnerability in MLflow's pyfunc extraction process allows for arbitrary file writes due to improper handling of tar archive entries. Specifically, the use of `tarfile.extractall` without path valid...