IBM revealed that it successfully completed field testing of fully homogeneous encryption. Full-homogeneous encryption (FHE) allows specific algebraic operations on redaction to be performed to be encrypted, just as the same operations are performed on clear text sending the results encrypted. This technology can be in encrypted data retrieval, comparison and other operations, the entire process does not need to decrypt the data.
The same-state encryption technology fundamentally solves the problem of confidentiality when entrusting data and its operations to third parties. Today’s files are usually encrypted during transmission and storage, decrypted when used, and have security vulnerabilities that FHE can resolve this issue. Of course FHE has its problem, which is the huge cost. FHE encrypted machine learning models require 40-50x compute power and 10-20x memory footprint compared to unencrypted. IBM conducted field tests of fully homogeneous encryption for a large U.S. bank and a large European bank, and the U.S. study has been published in 2019, while the European study remains confidential.