# ## What happened A short report on BleepingComputer says that researchers gained unauthorized control of the Hugging Face account hosting the text-to-speech model Kokoro after a lack of ownership verification on the platform.

How the platform works

Hugging Face is a public hub where developers upload and share pre-trained AI models. The article notes that Kokoro's page did not list a verified owner.

Why it matters

By accessing the account without permission, the researchers can influence how the model is trained, modified or distributed. The model is already employed by Google, NVIDIA, AMD, Intel and Amazon Web Services, so any changes could affect their products.

What remains unclear

The exact steps taken, whether the model's code was altered, and if safeguards were removed are unknown. No analysis of wider security risks was provided.