GPT-6 Astra and Claude Opus 5 Crack Unsolved Enigma Codes

Two frontier AI systems have autonomously cracked long-unsolved WWII Enigma ciphers, according to TechCrunch. Developer Carter Leffen prompted OpenAI's GPT-6 Astra to search a database and decode a message that had baffled researchers since 2005; the model independently handled the work by conducting archival research, identifying context clues, building an Enigma simulator, and recovering the plaintext message. Cybersecurity executive Jack Willis used Anthropic's Claude Opus 5 on September 21 to break a separate unbroken message, using an officer's name signature as the decryption key.

Frode Weierud, a retired electrical engineer and cryptology enthusiast who maintains the Crypto Cellar archive, validated Leffen's solution. 'GPT–6 Astra is behaving like a very professional cryptanalyst,' Weierud said. 'What it has achieved in two days would take a human researcher weeks or even months.'

Seven unsolved Enigma messages remain by Weierud's count, plus one where the plaintext is known but the code itself has not been recovered. The result lands amid a busy stretch for Anthropic in our tracker, including Claude's nine-loop Yang-Mills computation reported the same day, but the Enigma work is different in character: enthusiast validation on a narrow domain with a known-cleartext oracle, not peer-reviewed benchmark science.