OpenAI's Astra and Anthropic's Opus Crack Enigma

Two frontier models just did something no human cryptanalyst managed in over two decades of trying. Within days of each other, OpenAI's GPT-6 Astra and Anthropic's Claude Opus 5 each broke a German Army Enigma message that had sat unsolved since researchers first catalogued it as unbreakable.