AVID-2026-R1443

Description

Excessive directory permissions in MLflow leads to local privilege escalation when using spark_udf (CVE-2024-27134)

Details

Excessive directory permissions in MLflow leads to local privilege escalation when using spark_udf. This behavior can be exploited by a local attacker to gain elevated permissions by using a ToCToU attack. The issue is only relevant when the spark_udf() MLflow API is called.

Reason for inclusion in AVID: The CVE describes a software vulnerability in MLflow (a tool commonly used in ML pipelines) that enables local privilege escalation via spark_udf due to excessive directory permissions. This is a vulnerability affecting software components used to build/deploy/run AI systems, i.e., a general-purpose AI software supply chain issue. Evidence includes the CVE description and referenced PR, indicating a TOCTOU/permission-related exploit. The issue is not hardware-only and has clear security impacts (privilege escalation).

References

• NVD entry
• https://github.com/mlflow/mlflow/pull/10874

Affected or Relevant Artifacts

• Developer: Unknown
• Deployer: Unknown
• Artifact Details:

Impact

AVID Taxonomy Categorization

• Risk domains: Security
• SEP subcategories: S0100: Software Vulnerability
• Lifecycle stages: L06: Deployment

CVSS

CWE

Other information

• Report Type: Advisory
• Credits:
• Date Reported: 2024-11-25
• Version: 0.3.3
• AVID Entry