Non-Committing Encryption is Too Easy in the Random Oracle Model

Abstract

The non-committing encryption problem arises in the setting of adaptively secure cryptographic protocols, as the task of implementing secure channels. We prove that in the random oracle model, where the parties have oracle access to a uniformly random function, non-committing encryption can be implemented efficiently using any trapdoor permutation.

We also prove that no matter how the oracle is instantiated in practice the resulting scheme will never be non-committing, and we give a short discussion of the random oracle model in light of this.