Google has released an open set of libraries and utilities implementing a system of fully homomorphic encryption, allowing data to be processed in encrypted form, which does not appear in plain text at any stage of the computations. The toolkit enables the creation of programs for confidential computing that can work with data without decryption, including performing mathematical and simple string operations on encrypted data. The project code is written in C++ and is distributed under the Apache 2.0 license.
Unlike end-to-end encryption, homomorphic encryption not only protects data transmission but also allows data processing without decryption. Full homomorphism means the ability to perform addition and multiplication operations on encrypted data, from which any arbitrary calculations can be implemented. The output produces an encrypted result that would be equivalent to encrypting the result of analogous operations on the original data.
Working with data in homomorphic encryption involves the user encrypting the data and transmitting it for processing to a third-party service without revealing the keys. This service performs the requested computations and generates an encrypted result, having no ability to determine exactly what data it is working with. The user uses their keys to decrypt the provided data and obtains the result in plain text.
Applications of homomorphic encryption include the creation of cloud services for confidential computing, the implementation of electronic voting systems, the development of anonymized routing protocols, processing queries on encrypted data in databases, and confidential training of machine learning systems.
For example, homomorphic encryption will be useful in medical applications that can securely obtain confidential information from patients in encrypted form, allowing healthcare professionals to perform analytics and identify anomalies without decryption. Homomorphic encryption can also aid in research examining the link between diseases and specific genetic mutations, where analyzing thousands of genetic samples is required.
A distinctive feature of the released toolkit is the ability to create programs for processing encrypted data, using standard development practices in C++. With the provided transpiler, the C++ program is transformed into a special dialect FHE-C++ capable of working with encrypted data.
Source: opennet.ru
