zkGen: Policy-to-Circuit Transpiler

Jan Lauinger, Jens Ernstberger, Sebastian Steinhorst

Abstract: Modern privacy-enhancing technologies reach new forms of computation and data privacy according to a statement (e.g. private value $\gt \mathbf{100}$). However, beyond the statement expression, the description of a private computation circuit requires knowledge of security algorithms protecting private values. To keep the description of private and compliant computation circuits as close to the statement expression as possible, we introduce a new composable policy language called zkPolicy. Further, we introduce a policy transpiler, called zkGen, to decouple the complexity expressed via zkPolicy from the complexity of the underlying security algorithms. Our results show that, with zkPolicy, the description of compliant data provenance circuits can be reduced from 957 to 22 lines of code. And, with zkGen, we automate the generation and composition of private computation circuits to a minimum effort of configuring a zkPolicy.