Tools for Protecting Data and Function Privacy

Privacy is a fundamental right and ideally both users’ data and organizations’ intellectual property are protected. Privacy Enhancing Technologies (PETs) can protect data and function privacy. A mature PET to protect data is Multi-Party Computation (MPC).

Further PETs are Private Set Intersection (PSI) to compute functions on sets of elements, and Private Function Evaluation (PFE) to protect data and functions.

The main goal of PRIVTOOLS is to develop composable PETs and corresponding open-source tools that make them accessible to developers who are non-experts in cryptography. So far, MPC, PSI and PFE were studied mostly separately, and their composition to more complex functions is open.

We will develop composable and efficient PETs using various function representations.We achieve these objectives:1) Currently, each MPC framework uses a dedicated high-level language.

The objective for MPCTool is to give developers unified access to multiple MPC frameworks and function representations from multiple programming languages.

This allows technology transfer across MPC frameworks, compare their performance, and protects against vendor lock-in.2) For set operations, a developer must currently generate circuits for postprocessing elements in the intersection which is cumbersome.

To make PSI more accessible, PSITool will be the first tool for PSI that generates optimized PSI variant protocols from programs using common abstractions for sets.3) For private functions, a developer must currently generate a circuit that is evaluated with a PFE protocol. This is inefficient and often only parts of the function must be hidden.

PFETool will be a flexible tool for PFE that uses multiple function representations and automatically chooses efficient PFE protocols for subfunctions.The tools built in PRIVTOOLS also work with each other, e.g., PFETool can hide the set operations in PSITool. They will allow protecting data and functions in a large variety of applications.