What Real Systems Ask of Cryptography by Mr Zeyu Liu

Abstract

Online services and AI systems increasingly mediate how we communicate, analyze data, and make decisions. Yet privacy in these systems often still depends on trusting service providers with sensitive data. Furthermore, even end-to-end encryption primarily protects message contents; metadata, such as who communicates with whom and when, can remain exposed, enabling traffic analysis that reveals sensitive information. Meanwhile, cryptographic guarantees themselves may erode as computational capabilities evolve.

This talk therefore asks: What can cryptography promise real-world systems, and how can it keep those promises? I approach this question along three dimensions: scaling advanced cryptographic tools, identifying and addressing overlooked security requirements, and strengthening the post-quantum foundations on which such guarantees rely.

Biography

Zeyu Liu is a final-year PhD student at Yale University, advised by Prof. Ben Fisch. He received his M.S. from Columbia University, where he was advised by Prof. Tal Malkin and Prof. Eran Tromer, and his B.S. from UCLA.

His research focuses on cryptography, including fully homomorphic encryption, metadata privacy, and lattice-based cryptography. Much of his work has been published at leading cryptography conferences (e.g., CRYPTO, EUROCRYPT, and ASIACRYPT) and security conferences (e.g., IEEE S&P, ACM CCS, and USENIX Security).