Cryptography is a beautiful branch of mathematics which guarantees the art of secret keeping. In today’s world of big data, there is a serious conflict between utility and privacy. The prime example of this conflict could be found in field of medicine — the abundance of genetic data today makes it possible to imagine a future where medicines can be personalized, but developing this capability requires running algorithms on large scale genetic data which is highly sensitive, and may compromise the privacy of individuals. Another such example is the raging privacy debates pertaining to the biometric database in India (Aadhaar).

Resolving the conflict between utility and privacy is one of the primary challenges faced by cryptography. While traditionally designed to enable secure communication between two parties in the presence of an eavesdropper, modern cryptography has taken huge strides in expressiveness and generality, tackling paradoxical questions such as ‘if is it possible to run machine learning algorithms on encrypted data’ to ‘is it possible to prove that I know a secret without telling you what I know?’ to ‘is it possible to obfuscate computer programs so that they perform the intended functionality of the program but are guaranteed to reveal nothing about the inner workings of the program, even given its code?’ 

In a recent work to appear at Eurocrypt 2019, Dr Shweta Agrawal provided new approaches to the problem of constructing obfuscation and provided new candidate obfuscators based on novel mathematical hardness assumptions. This advances the state of the art in constructing obfuscation, and bringing it closer to resolution.