Artificial Intelligence Models Crack Unsolved World War II Enigma Messages
Experts have successfully decrypted long-unsolved World War II Enigma ciphers using OpenAI and Anthropic artificial intelligence models.
Following computing pioneer Alan Turing's success in cracking Nazi Germany's Enigma cipher during World War II, modern cryptanalysts have managed to read two previously unsolved Enigma messages using large language models from OpenAI and Anthropic.
AI Success in Turing's Footsteps
Although Alan Turing is best known for the test he designed to distinguish between machine and human intelligence, his most critical mission in life was breaking Enigma codes during World War II. Today, cryptanalysts are continuing this historical legacy with the help of modern large language models.
Archival Documents and Enigma Simulation
Developer Carter Leffen instructed OpenAI's Astra model to scan a message database to find and decrypt an unsolved message. Astra conducted archival research to find contextual clues, created an emulator of the Enigma machine, and recovered the plaintext of a message that had baffled researchers since 2005.
Expert Verification and Remaining Ciphers
Frode Weierud, administrator of the Crypto Cellar website, verified Leffen's solution. Additionally, cybersecurity executive Jack Willis broke a different unsolved message using Anthropic's Claude Opus 5 model. Weierud stated that there are still seven Enigma messages left in the archive waiting to be solved.