Paper 2023/1787

Updatable Privacy-Preserving Blueprints

Abstract

Privacy-preserving blueprint schemes (Kohlweiss et al., EUROCRYPT'23) offer a mechanism for safeguarding user's privacy while allowing for specific legitimate controls by a designated auditor agent. These schemes enable users to create escrows encrypting the result of evaluating a function $$, with $$ being publicly known, $$ a secret used during the auditor's key generation, and $$ the user's private input. Crucially, escrows only disclose the blueprinting result $$ to the designated auditor, even in cases where the auditor is fully compromised. The original definition and construction only support the evaluation of functions $$ on an input $$ provided by a single user. We address this limitation by introducing updatable privacy-preserving blueprint schemes (UPPB), which enhance the original notion with the ability for multiple users to non-interactively update the private user input $$ while blueprinting. Moreover, UPPBs contain a proof that $$ is the result of a sequence of valid updates, while revealing nothing else about the private inputs $$ of updates. As in the case of privacy-preserving blueprints, we first observe that UPPBs can be realized via a generic construction for arbitrary predicates $$ based on FHE and NIZKs. Our main result is UBlu, an efficient instantiation for a specific predicate comparing the values $$ and $$, where $$ is the cumulative sum of users' private inputs and $$ is a fixed private value provided by the auditor in the setup phase. This rather specific setting already finds interesting applications such as privacy-preserving anti-money laundering and location tracking, and can be extended to support more generic predicates. From the technical perspective, we devise a novel technique to keep the escrow size concise, independent of the number of updates, and reasonable for practical applications. We achieve this via a novel characterization of malleability for the algebraic NIZK by Couteau and Hartmann (CRYPTO’20) that allows for an additive update function.

Note: Added performance benchmarks & editorial.