Privacy Computing: Current Applications and Challenges in Finance, Healthcare, and Government

According to Gartner, privacy computing was identified as one of nine strategic technology trends for 2021, following 2020 being dubbed the "privacy computing year" and 2021 becoming the year of commercial deployment.

The 2021 (7th) China Internet Law Conference saw the China Academy of Information and Communications Technology release the "Data Value Release and Privacy Protection Computing Application Research Report (2021)", which concludes that privacy‑preserving computing technologies have begun exploratory applications in finance, healthcare, and government, but have not yet achieved large‑scale adoption.

This is the first domestic report focusing on privacy computing application research, addressing how data value, data privacy, and privacy‑preserving computation can unlock data value in finance, medical, and governmental scenarios, and providing references for technology adoption.

The report, jointly led by the China Academy and Ant Group, examines three major application scenarios.

In the financial scenario, privacy computing is deemed essential. The report notes that fraud risks cause billions of dollars in losses annually for the banking sector, and that privacy‑preserving computing platforms enable secure external big‑data fusion, enhancing risk‑control capabilities. A cited case shows a bank integrating data from over ten cross‑industry sources, reducing its non‑performing loan rate from 1.81% to 0.38% after joint modeling.

In the medical scenario, modern biomedical research, drug development, public health, and clinical applications increasingly rely on sharing electronic medical records, genomic, and imaging data. The report cites a top‑tier hospital that, using privacy‑preserving computation, performed GWAS analysis for ankylosing spondylitis without exposing raw individual genetic data, employing a secure federated learning framework that protected patient information throughout the process.

In the governmental scenario, the challenge is to unlock data value while safeguarding personal privacy. The report describes a collaboration between a technology company and a power utility that leveraged a self‑developed secure computing platform integrating secure multi‑party computation and federated learning to provide a power‑grid collaborative solution, building an electric‑vehicle load prediction model that supports grid planning and promotes the new‑energy vehicle industry.

Lin Meiyu, deputy director of the Academy’s Security Research Institute, emphasized that releasing data value requires clear definitions of data, data value, and data characteristics. Although pilot projects exist in finance, healthcare, and e‑government, privacy‑preserving computing remains in the early stage of large‑scale commercial adoption, facing challenges such as immature technology, incomplete solutions, and the need for coordinated policy, standardization, innovation, and digital security literacy.

Wei Tao, vice‑president of Ant Group, argued that privacy computing is an effective means to protect personal privacy while enabling data flow value, achieving the "computable but not identifiable" goal, which is key to balancing personal information protection with data‑driven industry development. Ant Group continues to explore privacy‑computing innovations across the entire data lifecycle, covering technologies such as secure multi‑party computation, trusted execution environments, zero‑knowledge proofs, and federated learning, and holds the highest number of global privacy‑technology patents (740). The report includes Ant Group’s case studies in finance and healthcare.

Key Takeaway: Click to read the original article and access the full report.