GPT-6 Cracks Napoleon's 217-Year-Old Code

🔥 What happened A developer with zero cryptography background pointed GPT-6 Astra at a coded 1809 Napoleon letter – and after six hours, the model cracked it. The letter revealed troop positions that had been unreadable for 217 years. 💡 Why it matters The AI started from just 33 known symbols and combined trial-and-error with French letter-frequency stats. This isn't a niche crypto tool – it's a generalist model casually doing cryptanalysis, a red flag for any legacy cipher still in use. ⚡ Our take When a hobbyist with a chat model breaks historical ciphers, it's not a PR stunt – it's a wake-up call for anyone confusing obscurity with security.