|Google Open Sources Fully Homomorphic Encryption Transpiler|
|Written by Kay Ewbank|
|Thursday, 17 June 2021|
Google has open sourced a fully homomorpthic encryption (FHE) transpiler. FHE provides a way to carry out computations on encrypted data without it needing to be decrypted.
For normal encryption techniques, data can be encrypted at rest and in transit, but if you want to use it, it has to be decrypted, at which point it is exposed and vulnerable to attack. Fully Homomorphic Encryption avoids this problem by providing a way to work with the encrypted data.
The transpiler will let developers use encrypted data without needing or being able to access any personally identifiable information. By using FHE, encrypted data can move between servers and be processed without being decrypted. Google’s transpiler means developers can write apps that run on encrypted data, and the transpiler transforms the code into a version that can run on encrypted data. FHE can also be used to train machine learning models on sensitive data in a private manner.
The FHE C++ Transpiler is a general purpose library that converts C++ into FHE-C++ that works on encrypted input.
Discussing the new release, Miguel Guevara, Product Manager, Privacy and Data Protection Office at Google said the introduction is part of Google's continuing work on making Google's our products private by design. He drew parallels with Google's Federated Learning, the technology used to improve models on devices without sending users' raw data to Google servers. Google Assistant uses federated learning to improve “Hey Google.” Guevara also referenced Google's move to make its differential privacy library freely available to any organization or developer. The library is an advanced anonymization technology that enables developers to learn from their data privately.
The new release is available on GitHub.
or email your comment to: email@example.com
|Last Updated ( Saturday, 19 June 2021 )|