Driving Innovation with Differential Privacy

Oasis Labs launches an early-stage project with The BMW Group to build the next generation of data privacy technology.

Published in
3 min readJan 29, 2021

--

Introduction

Data is the lifeblood of modern companies. It powers products, drives decisions, and closes business agreements — yet with more data comes an increased risk for exposure, accidental leaks, and privacy violations. Recent regulations such as GDPR and CCPA have taken steps to protect individual privacy, but they also add a significant new barrier to data use that only new technology can solve.

Oasis Labs offers privacy and confidential compute technologies that can enable responsible data use, where individuals retain control of their data as self-sovereign owners of the information they produce, and corporations use this data with consent from individuals with technologies that preserve individual privacy and confidentiality. The solution leverages a powerful new brand of technology called differential privacy to provide programmatic privacy guarantees for access to a particular data set.

Oasis Labs is working with the BMW Group on an early-stage project to create innovative privacy solutions that leverage differential privacy and set a new standard for responsible data use in the automotive industry.

A Case for Differential Privacy

Imagine you have a database of employee salaries. Assume that a query that you permit on the database is the average salary of the employees in the database. If Bob knows the number of employees in the company and runs this query before and after Chloe joins the organization, then Bob can calculate Chloe’s salary as shown below,

  1. Bob knows the number k of employees in his company
  2. Bob runs an average salary query and gets N
  3. Chloe joins his company
  4. Bob runs the average salary query and gets M
  5. Chloe’s salary = M(k + 1) — Nk
Figure 1: Differential privacy ensures that query results maintain privacy
Figure 2: The differential privacy mechanism

Differential privacy is a technique that guarantees that the results of statistical queries cannot be used to glean any information about specific individuals or more broadly access specific rows in a database. Information can only be accessed in the aggregate. The Oasis solution for differential privacy works for SQL databases and is based on query rewriting. The guarantees that are provided and the mechanism that is used are shown in Figures 1 and 2. One of the advantages of using a query rewriting approach is that any database that supports a SQL dialect that includes the mathematical functions abs, random, ln, and sign can be used as the backend database. The mechanism renders queries intrinsically private. Once rewritten, the query can be submitted to the database and the results are differentially private.

Providing Differentially Private Solutions for The BMW Group

In a new, early-stage, Oasis Labs is working with the BMW Group to test applications of differential privacy in their internal systems. This solution enables both internal teams and external partners to access data while remaining compliant and protecting user privacy. All accesses can be persisted on a ledger to enable consent-based audit, and all access policies can be checked with high integrity by the Oasis Labs platform prior to running queries and returning results. The integration builds on the model’s already robust privacy and security infrastructure, bringing the latest in privacy technology to their stack.

Towards a future of responsible data use

The joint BMW Group-Oasis Labs effort seeks to explore the latest in privacy technology. For Oasis Labs, this integration is just another example of how we’re working with industry leaders to bring new data privacy practices and technologies into our modern world, and bringing us one step closer to a responsible data economy.

--

--