Paper 2025/766

Unbiasable Verifiable Random Functions from Generic Assumptions

Abstract

We present conceptually simple and practically competitive constructions of verifiable random functions (VRF) that fulfill strong notions of unbiasability recently introduced by Giunta and Stewart. VRFs with such strong properties were previously only known in the random oracle model or from the decisional Diffie–Hellman assumption with preprocessing. In contrast, our constructions are based on generic assumptions and are thus the first to be plausibly post-quantum secure in the standard model (without any setup). Moreover, our transformation preserves useful properties of the underlying VRF such as aggregatability, (a form of) key-homomorphism, small entropy loss, and computability in ${\mathsf{NC}}^1$; and it even yields a symmetric unbiasable VRF whose pseudorandomness holds even when the input and the key are swapped. To underscore the importance of a provably unbiasability in the standard model, we showcase a potential security weakness in the folklore VUF-then-Hash construction. Lastly, we discuss and remedy several issues regarding the definition of unbiasability, and outline a path towards a lattice-based instantiation of VRFs.

Note: Reorganization, add symmetric VRF.