Skip links

Alibaba affiliate Ant Group open sources its privacy software and a ‘Secure Processing Unit’

Alibaba’s financial services affiliate, Ant Group, has open sourced its “privacy-preserving Computation Framework.”

The goal of the release, according to an Ant announcement, is “to make the technologies more accessible to global developers and speed up the framework’s application.”

The Framework, called SecretFlow, can be found on both GitHub and China’s analog Gitee. In the repos you’ll find code for:

  1. A “Secure Processing Unit” that offers a “provable, measurable secure computation device, which provides computation ability while keeping your private data protected”
  2. A homomorphic encryption processing unit called “heu”
  3. Secureflow, billed as “a unified framework for privacy-preserving data analysis and machine learning”
  4. A portable simplest oblivious transfer library called “simplest-OT”
  5. YASL (Yet Another Simple library), a C++ library containing modules and utilities that Secretflow requires

SecretFlow, which is designed to promote privacy during data analysis and machine learning efforts, was under development for six years and has been used internally at Ant Group already, as well as a few external organizations.

Ant Group’s privacy computer general manager, Lei Wang, billed the endeavor as a zero-cost shortcut to privacy for developers.

The Alibaba affiliate is no stranger to open-sourcing – it placed its low-latency service registry SOFARegistry on GitHub in 2019 and its interface design language Ant Design in 2015.

The financial services company also announced that it was investing over $600,000 in a newly launched fund. The endeavor, titled the CCF-Ant Group Privacy Computing Special Fund, is in collaboration with professional organization China Computer Federation.

The cash will go to researching privacy computing, including post-quantum multiparty computation, and will be available worldwide.

Ant’s announcement of the code release reminded world+dog it topped the list of patent applications for privacy-preserving computation technologies in 2022 and therefore fancies itself an expert in a field it expects will continue growing. ®