Arbitrary Code Execution Vulnerability in MLflow Platform

CVE-2024-37059

Summary

The MLflow platform contains a vulnerability where deserialization of untrusted data can occur in versions 0.5.0 and above. This vulnerability enables an attacker to upload a malicious PyTorch model, which, when interacted with, can execute arbitrary code on an end user's system. This poses significant risks to system integrity and confidentiality, as malicious code can compromise user data and potentially lead to larger security breaches. Implementing security best practices and promptly updating affected versions can help mitigate this threat.

References