Breaking: the Enigma code just got cracked again. This time, no Alan Turing, no Bombe machine.
Everyone knows Turing for the test named after him, the one about telling humans and machines apart. But the real test of his life was breaking Enigma, the cipher Nazi Germany used to encrypt wartime communications.
Now, two cryptanalysts say they've decoded two long-unsolved Enigma messages that had stumped historians and codebreakers for decades.
How? They used LLMs like Astra from OpenAI and Opus from Anthropic. Models built to chat, write, and reason in natural language, not to break military ciphers.
That's what makes this wild. Turing's team built an entire purpose-built machine, the Bombe, to do one job: crack Enigma. Here, general-purpose AI models solved it as a side effect of just being good at pattern recognition and language.
If a chatbot can casually crack a puzzle that stumped experts for 80 years, the question isn't 'is AI smart' anymore. It's 'what else is it quietly capable of that we haven't even asked it to try?'
🔗 Original Source & Reference: https://techcrunch.com/2026/09/25/astra-and-opus-just-passed-turings-other-test/
Published automatically via FeedMind AI Content Pipeline.
Top comments (0)