Contrary to popular assumption, most "zero-knowledge” systems today aren’t private. Drawing from decades of research, @mvenkita (CEO of @ligero_inc) breaks down in this @WPReadingClub session why the majority of proof systems in use - including many STARK-based zkVMs - deliver verifiable compute but fall short of true cryptographic privacy. Muthu explains how Ligero and Ligetron change that equation: • Designed for time and memory efficiency, they can generate proofs even on memory-constrained devices like mobile phones. • Built on MPC-in-the-head constructions, they guarantee provable privacy where many current systems do not. • Already adopted by @googlecloud for mobile identity verification, showing real-world viability. The discussion spans the academic foundations of ZK, the evolution from theory to implementation, and the pressing need for proofs that protect sensitive data while still meeting compliance requirements. Muthu highlights emerging applications - privacy pools, zk-validiums, and on-device identity proofs - that could define the future of enterprise and regulatory adoption, and touches on experiments proving large ML models (like LLaMA) with ZK, pointing to what comes next for privacy-preserving AI. This is a direct look at the gap between what we think ZK does today, and what’s required for a world where privacy, compliance, and scalability truly coexist. Also available on YouTube: