Zero-Knowledge Proofs Do Not Solve the Privacy-Trust Problem of Attribute-Based Credentials: What if Alice Is Evil?

Abstract

Zero-knowledge schemes have recently become a popular attempt to offer users privacy in an attribute-based credential system. In this article, we do not contest the mathematics of these schemes; we assume it is logically sound. Instead, we draw attention to the trade-off that is made when employing cryptography instead of trusted parties to protect user privacy. We assert that, for these approaches to create the trust required by credential verifiers, they must introduce mechanisms that limit their utility and create significant privacy risk to the user that cuts against data minimization goals. Greater trust must be placed in the shelf life of cryptography to prevent the user from being unwantonly correlated than alternative approaches. Just as we would discourage storing encrypted private data on public blockchains, we discourage this approach here. Lastly, this article introduces the concept of a trusted witness which provides privacy for honest users and solves the privacy-trust problem without the disadvantages of the zero-knowledge approach.