In this episode of Science Will Make You Happy, Dan Bogdanov, Chief Scientific Officer at Cybernetica, is joined by Riivo Talviste, product owner for Sharemind MPC and security engineer at Cybernetica, to explain secure multi-party computation (MPC) — one of privacy technology's most powerful, and most misunderstood, tools.
They cover:
- What MPC actually means, in plain English
- Real MPC projects: a confidential financial benchmarking survey for Estonian ICT companies, and satellite collision prediction between countries that won't share their orbital data
- Why MPC needs less trust than a normal client–server system, replacing a single "trusted third party" with several mutually distrusting computing parties
- How MPC compares with (fully) homomorphic encryption and trusted execution environments (TEEs) — and why combining MPC with TEEs adds extra guarantees
- 20 years of Sharemind: from a 2006 university thesis to a commercial MPC platform
- Inside JOCONDE, a new Eurostat-funded project bringing MPC-as-a-service to official statistics across the EU
- Emerging use cases: healthcare, cyber-threat-intelligence sharing, anti-money-laundering, and fraud detection for tax authorities and the UK Cabinet Office
- Whether cryptographically secured computation counts as "anonymous" data under EU law
- Whether an AI coding agent could build an MPC system today — and what still can't be cloned
Prefer audio? Also available on Spotify and Apple Podcasts.
