OpenAI GPT–6 Astra breaks Enigma message that has resisted solution since 2005

OpenAI's GPT-6 Astra AI system independently broke a German Army Enigma message from July 1941 that had resisted decryption attempts for over two decades. The AI autonomously analyzed historical encrypted messages, identified the most promising target (message MVUEH), developed its own Enigma simulation software, and used linguistic cribs to successfully recover the correct decryption key and plaintext. The breakthrough was complicated by transcription errors in the original ciphertext and a rare left-hand wheel turnover in the Enigma machine, factors that likely prevented human codebreakers from solving it earlier.