Date and time: Nov 13, 2025, 1-2:30pm
Location: Rice 340
Title: Lower Bounds on IPFE from All-or-Nothing Encryptions
Abstract:
Functional encryptions (FE) provide fine-grained access control over encrypted data. The class of functions supported by an FE scheme can vary widely, from all polynomial-size circuits to much simpler functions, such as inner products. FE for all circuits is extremely powerful and can be existentially equivalent to indistinguishability obfuscation (iO). In contrast, FE for inner products (IPFE) can be based on standard assumptions such as DDH and LWE.
In a very recent work, Hajiabadi, Langerhr, O’Neil, and Wang showed that even the secret-key IPFE is impossible in the random oracle model (ROM). Following their results, we focus on the feasibility of this minimal but non-trivial class of functional encryption, IPFE. We proved that even several fancy encryption primitives including IBE, ABE, FHE, and PE are insufficient for IPFE. Our impossibility results suggest the intrinsic difference between FE and these fancy encryptions: FE enables partial decryption, whereas others share an “all-or-nothing” nature.
In this talk, I will briefly review IPFE and black-box separation techniques. I will then give an overview of our separation between IPFE and FHE, focusing on how the attack framework used to separate IPFE from ROM can be adapted and extended to rule out constructions from FHE.
This is a joint work with Jinye He and Wei-Kai Lin.