Unsloth Studio Flaw Turns Routine Model Inspection Into Code Execution

A patched Unsloth Studio vulnerability allows malicious AI models to execute arbitrary Python code during inspection, via the trust_remote_code setting.