GPT-6 Astra Reportedly Cracked an Unsolved 1941 Enigma Message On Its Own
A viral demonstration claims the model autonomously searched archives, built a simulator, and recovered plaintext from a German Army cipher that had gone unbroken for over 80 years.
If verified, it is the kind of result that makes an abstract capability tangible. According to @rohanpaul_ai, OpenAI's GPT-6 Astra cracked a previously unsolved 1941 German Army Enigma message — not by being handed the tools, but by autonomously searching historical archives, constructing its own Enigma simulator, and running the process end to end until it recovered readable plaintext.
Unlock the full briefing
Get every story in today's briefing, the full archive, and the daily AI intelligence brief.
All stories today
Full archive
Daily brief
Cancel anytime. Payments powered by Stripe.