Google’s AI Are Sending Encrypted Messages to One Another That No One Can Decipher

Getty Creative

You have to admit; it sounds a little worrying, right? Google’s multiple AI’s that make up a large part of the Google Brain Project have been taught not only how to create their own encrypted messages but to also use them with one another when no one else can read them.

Alice, Bob, and Eve are three neural networks that were created as part of the Google Brain project in an attempt to get to the bottom of deep learning techniques. Every day they are in operation, they are getting smarter and smarter, and now it seems they have just mastered encryption.

During testing, the task that Alice was set was to create a simple form of encryption and work alongside Bob to devise a key made up of an agreed set of numbers.  Bob was able to read the message fine using the key they created together.  Simultaneously, Eve decides to try and intercept the message and decipher it.  But, Eve failed and was unable to break the encryption.

However, the great things about these machines are that they learn continuously, so although they were performing pretty poor, to begin with when it came to the secret message passing, Alice at least saw remarkable improvements in its capabilities and over time would perform even better.  In terms of digital security, this will improve significantly when theses AI systems are let loose and at the very least will cause hackers some problems they would rather do without.

More News To Read