Leveraging Privacy-Enhancing Technology to Better Serve the United States’ Public

While the public views the federal government as a monolith, in truth United States departments and agencies often function independently. Due to privacy regulations and statutory reasons, they often cannot share data about individuals with each other. Yet, such data collaboration would facilitate the development of artificial intelligence models to significantly improve the provision of social services. We argue that a particular privacy-enhancing technology (split learning operating over vertically partitioned data), together with data governance through metadata supply chains, can be part of larger sociotechnical systems that can improve life for the most vulnerable among us.