"Cryptographic Agents" is a new framework that unifies various cryptographic objects, similarly to how the Universal Composition framework unifies various multi-party computation tasks. Cryptographic Agents can model existing and new versions of various forms of obfuscation, functional-encryption, fully-homomorphic encryption and even certain hardness assumptions related to primitives like multi-linear maps. The framework uses a new definition that does not rely on simulation, and hence side-steps typical impossibility results. It also admits a composition theorem, which can be used to derive new constructions.

I shall present the basic framework (appeared at Eurocrypt'15) and some recent results.

Based on joint work with Shweta Agrawal, Shashank Agrawal and Ching-Hua Yu.