Hu Yoshida

AI has been demonstrated to break the Enigma code in 10 minutes

Blog Post created by Hu Yoshida on Jan 29, 2018

Several years ago, I posted a blog on how Alan Turing was able to break the Enigma code, the secret language used by the Nazis in WWII that was originally thought to be unbreakable. With a polyalphabetic substitution cipher that was many times the length of the longest message, a 4-rotor scrambler, and with 10 leads on to a plug-board that set the number of ways that the pairs of letter could be interchanged, it had a possible 151 trillion permutations and combinations to decipher.

Enigma.png

Alan Turing started work on the Enigma code in 1939 and was able to break some of the code as early as 1940 with the help of a team of mathematicians and logicians at Bletchley Park in the UK during World War II and with prior information from Polish intelligence who were able to break earlier Enigma code. Each branch of the German armed services managed and maintained their own Enigma systems so some were easier to decode then others. The Naval Enigma systems were the hardest to break consistently since the plug-board setting was changed on a daily basis and they were the most disciplined in its use. It usually took two days to decipher a new key setting which was useless when the keys were changed every 24 hours. Solving the Naval Enigma code was critical for the D day invasion in July 1944. Breaking the Enigma code is credited for shortening World War II by at least 2 years.

 

Turing developed a machine that could do a brute force decryption that took several days but this could be shortened if they could uncover certain phrases that might be repeated with different encryption keys. This is where the mathematicians and logicians helped the most. Turing looked for people with a creative imagination, a well-developed critical faculty, and a habit of meticulousness. Skill at solving crossword puzzles was famously tested in recruiting some cryptanalysts. These are personal traits which are still highly desired as we look for analysts today.

 

An AI company recently demonstrated how the use of AI and cloud servers could break the Enigma code in just over 10 minutes Unlike the traditional method of cracking the code, the AI was trained to look for German language, and then work out the statistical probability of the sentence decrypted being the accurate original based on how ‘German’ it was, using 2,000 cloud servers to do the calculations. Ai took a different approach to solving an encryption challenge that can be used in solving many other challenges like finding cures for different cancers.

 

What is disturbing is that an AI system like this could crack a 20 character password in less than 20 min.

That’s why we need to look beyond encryption to biometrics for systems of authentication. See my previous blog on Biometrics.

Outcomes