Overview

Data has three primary states - at rest, in motion, and in computation. When it comes to data security, we have solved data security at rest and in motion through encryption. But data computation still is mostly done today in plaintext leading to data leaks and privacy concerns.

FHE is the most common cryptographic primitive used for confidential data computation but it is too slow to be meaningfully used for most applications today. Recognising this limitation, OpenVector is building a new confidential compute primitive that is 100 times faster than FHE. The first use case it enables uniquely is confidential AI. OpenVector's vision is to build the Compute Layer Security (CLS) protocol to make every computation encrypted by default.