Unsloth Studio Flaw Turns Routine Model Inspection Into Code Execution

Summary

A vulnerability in Unsloth Studio, a tool for optimizing AI models, has been patched. The flaw, identified as CVE-2024-41104, allowed malicious AI models to execute arbitrary Python code on a user's system when the `trust_remote_code` setting was enabled during model inspection. This could have turned a routine model check into a security risk.

Indicators of Compromise

• cve — CVE-2024-41104

Entities