How Google's AI taught itself to create its own encryption

As machine learning becomes ubiquitous, robots will be tasked with handling increasingly more sensitive and private data. In order to help protect this personal information, computer scientists at Google have developed neural networks that teach themselves how to encrypt the information they process. A team from Google Brain, the organisation's deep learning research project, taught neural networks how to encrypt and decrypt messages. In a research paper published online the scientists created three neural networks: Alice, Bob, and Eve. Each was assigned its own job.