Time to Rethink: Trust Brokerage using Trusted Execution Environments

Mining and analysis of digital data has the potential to provide improved quality of life and offer even life-saving insights. However, loss of privacy or secret information would be detrimental to these goals and inhibit widespread application. Traditional data protection measures tend to result in the formation of data silos, severely limiting the scope and yield of ``Big Data''. Technology such as privacy-preserving multi-party computation~(MPC) and data de-identification can break these silos enabling privacy-preserving computation. However, currently available de-identification schemes tend to suffer from privacy/utility trade-offs, and MPC has found deployment only in niche applications. As the assurance and availability of hardware-based Trusted Execution Environments~(TEEs) is increasing, we propose an alternative direction of using TEEs as ``neutral'' environments for efficient yet secure multi-party computation. To this end, we survey the current state of the art, propose a generic initial solution architecture and identify remaining challenges.