Indistinguishability Obfuscation and Functional Encryption for all circuits
The goal of secure program obfuscation is to make a program ``unintelligible'' while preserving functionality. Unfortunately, 12 years ago, Barak et al. showed that the most natural simulation-based formulation of this problem is impossible to achieve. However, Barak et al. also proposed the alternative notion of indistinguishability obfuscation, which requires that given any two equivalent circuits C and C' of similar size, the obfuscations of C and C' should be computationally indistinguishable.
While indistinguishability obfuscation for general circuits has been a tantalizing open question, it has also been an enigma: It is not clear what, exactly, an indistinguishability obfuscator must hide about the circuit it transforms. Therefore, it has not been clear how useful such an obfuscator would be.
In this talk, we present a construction of indistinguishability obfuscators for all circuits using ideal lattices, and give new techniques to leverage the power of indistinguishability obfuscation for applications. In particular, we use indistinguishability obfuscation as a critical tool to resolve the open question of functional encryption for general circuits:
In functional encryption, ciphertexts encrypt inputs x and keys are issued for circuits C. Using the key SK_C to decrypt an encryption of x yields the value C(x), but does not reveal anything else about x. Furthermore, no collusion of secret key holders should be able to learn anything more than the union of what they can each learn individually. The functional encryption scheme we construct also enjoys succinct ciphertexts, which enables several other applications.
If time permits, I will also talk about additional applications of indistinguishability obfuscation, and open problems in obfuscation.
This talk will be based on joint works with Sanjam Garg, Craig Gentry, Shai Halevi, Mariana Raykova, and Brent Waters.