Private LLM Inference with Homomorphic Encryption
Lawrence Lim, Divyakant Agrawal, Amr El Abbadi · Proceedings of the VLDB Endowment · 2026
This tutorial introduces the emerging research area of private large language model (LLM) inference using Fully Homomorphic Encryption (FHE). It covers the basics of homomorphic encryption, key architectural details of transformer-based LLMs, and existing techniques for enabling end-to-end encrypted LLM inference. Finally, the tutorial concludes with open research questions and future directions in this rapidly evolving field.