AI models decode two long-unsolved Enigma messages
Two cryptanalysts say large language models from OpenAI and Anthropic have decoded separate messages sent using Nazi Germany’s Enigma machines during World War II. A small number of archival messages had remained unresolved, often because of transcription mistakes or errors made by the original operators.
Developer Carter Leffen asked OpenAI’s Astra to find an unsolved message in a database maintained by cryptology researcher Frode Weierud and decode it. Astra conducted archival research, identified contextual clues, built an Enigma simulator, and recovered the plaintext of a message that had challenged researchers since 2005. Weierud later validated the result and praised the model’s work. Astra’s logs also refer to messages in a private collection, although it is unclear whether the model accessed those materials.
On September 21, cryptanalyst Jack Willis told Weierud that Anthropic’s Claude Opus 5 had solved another message. Willis supplied more guidance, including the known signature of an officer’s name. Weierud says seven Enigma messages remain undecoded, plus one whose plaintext is known but whose code has not been solved.
Written by Sakhanda Wire AI from reporting by TechCrunch. Facts, quotes and full context are in the original article below.