Amateurs use large language models to break an Enigma message unsolved for 85 years

16 October 2026

Non-experts are using large language models to break historical codes, harnessing their ability to rapidly deploy decryption algorithms while also searching the internet for extra historical documents. In September, Carter Leffen, a product development coach at Bloomberg in New York City who has no cryptography experience, asked the model GPT-6 Astra, developed by OpenAI, to scour Crypto Cellar Research, a website of unsolved historical ciphers run by independent cryptographer Frode Weierud in Oslo, and to decode a message encrypted by Enigma, the machine used by the German armed forces during the Second World War. Enigma's rotors and plugs rerouted electrical signals to encrypt the alphabet in up to 103 sextillion possible ways. The model chose a message sent on 10 July 1941 that was 82 characters long, wrote code to simulate an Enigma machine, and in ten hours tested more than four billion settings. Days later Carter Church, a staff engineer at SentinelOne in Mountain View, California, used the same model on an unsolved cipher sent by an Italian official to one of Napoleon's generals in 1809, listed on Cryptiana, a website compiled by Japanese patent engineer and hobby cryptographer Satoshi Tomokiyo. Using a common strategy called simulated annealing, the model hit on a successful key in about six hours, and the decoded message described troop locations. Cryptography specialists are being overwhelmed by requests to validate AI solutions, many of which are incorrect; Elonka Dunin, a retired game developer who has written books on cryptography, said there are many false positives from people who do not understand cryptanalysis.