Cryptanalysts say OpenAI and Anthropic models cracked two long-unsolved Enigma messages
One developer had GPT-6 Astra find, research and decode a message that had defeated researchers since 2005.
Two cryptanalysts say they have decoded two long-unbroken Enigma messages using large language models from OpenAI and Anthropic, TechCrunch reports. A small number of archived Enigma messages have resisted decryption, usually because of transcription errors or mistakes by the original operators. Developer Carter Leffen told OpenAI's Astra model to search a database of Enigma messages for one that remained unbroken and to decode it. The model did its own archival research, identified contextual clues, built a simulator of the Enigma machine and recovered the plaintext of a message that had baffled researchers since 2005. Leffen also used Astra to build an interactive website explaining the problem. Frode Weierud, a retired electrical engineer who maintains the Crypto Cellar website and its database of messages, validated Leffen's solution last week and said the result left him in 'awe.' The second decryption involved an Anthropic model, according to the report. The work is a small historical result rather than a security finding, but it shows models carrying out multi-step research that combines archival digging, tool-building and verification by an outside expert.