Zama's research on the application of Fully Homomorphic Encryption (FHE) to Large Language Models is now featured on @huggingface. Their Machine Learning team recently discovered that FHE has the potential to safeguard both user privacy and the Intellectual Property of the model.