Apple’s open-source “swift-homomorphic-encryption” package revolutionizes privacy in cloud computing. It allows computations on encrypted data without decryption, safeguarding user information. Used in iOS 18 Live Caller ID Lookup, this technology enables secure data processing in various fields, including healthcare, finance, and AI.
Apple has introduced a new open-source package for its programming language Swift, called “swift-homomorphic-encryption.” This innovative technology ushers in a new era of privacy-preserving cloud computing with the power of homomorphic encryption (HE).
Traditional encryption methods render data unreadable until it’s decrypted. While this protects data confidentiality, it often hinders analysis. Homomorphic encryption, however, offers a revolutionary approach. This cryptographic technique allows computations to be performed on encrypted data itself, eliminating the need for decryption and subsequent risk of exposure.
In this approach, clients send encrypted data to a server, which operates on it and returns a decryptable result. The server never decrypts the original data or has access to the decryption key. This approach offers new opportunities for cloud services while protecting user data privacy and security.
Apple has open-sourced its HE library, which it uses in its own devices, available under Apache 2.0 license. The “swift-homomorphic-encryption” package, provides developers with the tools to integrate homomorphic encryption functionalities into their applications.
Notably, Apple has demonstrated the package’s application in its iOS 18 Live Caller ID Lookup feature, which provides caller ID and spam-blocking services. Using HE, users can send an encrypted query to a server to get information about a phone number without “the server knowing the specific phone number in the request,” Apple explained.
Live Caller ID Lookup uses Private Information Retrieval (PIR), a private key-value database lookup, to retrieve a private keyword without the server learning it. A trivial implementation of PIR is sending the entire database to the client for local processing, but this is only feasible for small databases with infrequent updates.
However, when this implementation uses homomorphic encryption, it only needs to sync a small amount of database metadata with the client, which changes infrequently, allowing efficient handling of large databases with high update volumes.
The swift homomorphic encryption package uses the Hummingbird HTTP framework for cross-platform support, a Benchmark Library for performance benchmarking, and Swift Crypto for low-level cryptography primitives. It uses the Brakerski-Fan-Vercauteren (BFV) HE scheme for post-quantum 128-bit security, ensuring secure computation on encrypted data and protection against classical and potential quantum attacks.
Pascal Paillier, Co-Founder and CTP of ZAMA AI, explains how Fully Homomorphic Encryption functions:
Homomorphic encryption has the potential to revolutionize data handling in various fields like healthcare, finance, and national security. It can enable secure machine learning on encrypted data, leading to breakthroughs in artificial intelligence. With continued development, homomorphic encryption could transform the way we interact with data in the digital age.
