CVE-2024-24590: 
ClearML Server vulnerability analysis and mitigation

A deserialization vulnerability was discovered in Allegro AI's ClearML platform client SDK versions 0.17.0 to 1.14.2. The vulnerability was identified and disclosed by HiddenLayer's SAI team, with the CVE being assigned on January 25, 2024 (CVE Mitre, HiddenLayer Research).

The vulnerability involves the insecure handling of pickle files in the ClearML Python package. When a user calls the get method within the Artifact class to download and load a file into memory, malicious pickle files can be deserialized on their system, leading to arbitrary code execution. The vulnerability is classified as CWE-502, relating to deserialization of untrusted data (NVD CNA).

The vulnerability enables maliciously uploaded artifacts to execute arbitrary code on an end user's system when interacted with. This could potentially lead to complete system compromise, allowing attackers to run malicious code in the context of the user accessing the artifact (HiddenLayer Research).

The vulnerability was responsibly disclosed to ClearML, and the team worked to resolve the issues within the 90-day disclosure window. Users should upgrade to versions newer than 1.14.2 of the ClearML client SDK to mitigate this vulnerability (HiddenLayer Research).

The discovery of this vulnerability, along with five others in the ClearML platform, has highlighted the need for increased security scrutiny in MLOps tools. The responsible disclosure process was praised for demonstrating good working relationships between security teams and product developers (HiddenLayer Research).