Programming Applications Suitable for Secure Multiparty Computation Based on Trusted Execution Environments

Abstract

Secure Multiparty Computation enables secure processing of the data from multiple users without ever exposing the data or the computation code of one user to the other parties or computation hardware provider. Such an approach enables new applications and better data computation results through secure and private data sharing and making richer datasets for the analysis. This paper explains how Secure Multiparty Computation is made possible by recent development in the computer processor industry and the popularization of the Trusted Execution Environment functionality. Four problems that need to be addressed are highlighted and solutions for two of them are presented. Design for one of the key components of a Confidential Computing System for Artificial Intelligence is provided. It is the RESTful web application implemented in the Python programming language using Flask framework, which is running on a secure virtual machine.