GPT-SoVITS-WebUI Unsafe Deserialization Vulnerability Leading to Remote Code Execution

A vulnerability allowing unsafe deserialization, which can lead to remote code execution, has been identified in GPT-SoVITS-WebUI versions through 20250228v3. The issue arises in the process_ckpt.py file, where the SoVITS_dropdown variable accepts user input and passes it to the load_sovits_new function. This function uses the input to load a model with torch.load, creating a risk of unsafe deserialization.

Impact

Exploitation of this vulnerability allows for remote code execution on the server where GPT-SoVITS-WebUI is running.

Reproduction

To reproduce this vulnerability, upload a malicious model file that includes executable code to a location accessible by the GPT-SoVITS-WebUI. Then, select this file in the SoVITS dropdown menu. The application will load the model using torch.load, executing the embedded code and demonstrating the remote code execution vulnerability.