Two Cryptanalysts Say OpenAI and Anthropic Models Broke Unbroken Enigma Messages
Two cryptanalysts say they used OpenAI's Astra and Anthropic's Claude Opus 5 to decode Enigma messages that had gone unbroken for decades, with the archivist who runs Crypto Cellar validating one solution.
Frode Weirerud, a retired electrical engineer with a lifelong interest in cryptology who runs the website Crypto Cellar, validated the first of the two solutions last week. According to TechCrunch, the result left him in "awe."
The solution came from Carter Leffen, a developer, who told OpenAI's newest model, Astra, to search a database of Enigma messages for one that had never been broken and decode it. The model carried out its own archival research, found context clues, built a simulator of the Enigma machine and recovered the plaintext of a message that had baffled researchers since 2005.
In remarks quoted by TechCrunch, Weirerud wrote that "GPT-6 Astra is behaving like a very professional cryptanalyst and archive researcher." He said the work took the model two days, where a human researcher would need weeks or even months, and that he had personally spent several weeks researching the Bundesarchiv files the model cited.
The model's logs discuss archived messages held in a "private collection" that Weirerud does not host. He said he is still unsure whether Astra accessed those messages, and speculated that they may have been shared by another researcher somewhere online, or that the model was able to reach the German government's public archives.
A second cryptanalyst, Jack Willis, contacted Weirerud on September 21 and said he had used Anthropic's Claude Opus 5 to break a different unbroken message. Willis gave the model considerably more guidance than Leffen gave Astra, and Claude ultimately used the known signature of a particular officer's name to break the text.
Enigma machines were used by Nazi Germany during World War II, and Alan Turing, best known for the test of whether machine and human intelligence can be distinguished, led the effort against the cipher. Turing and his team of cryptanalysts built an early computer, the Bombe, which allowed Britain to read Enigma messages during the war. A handful of archival messages were never solved, usually because of mistranscription or errors by the original encoders.
Weirerud says seven unbroken Enigma messages remain, along with one message whose plaintext is known but whose code has still not been recovered.
Editor's Summary
Two cryptanalysts say large language models from OpenAI and Anthropic solved Enigma messages that had resisted decoding, and Crypto Cellar's maintainer Frode Weirerud has validated one of the results. Weirerud said the models worked far faster than human researchers, and he noted that seven unbroken Enigma messages remain in the archive.