Claim: GPT-6 Astra cracked a 217-year-old Napoleon cipher in six hours
A story post says a user uploaded a photo of an 1809 coded letter and Astra read 1,300 handwritten symbols, wrote its own code to break the cipher, translated the message and completed a sentence in Napoleon's memoir, with a cipher historian verifying the result.
this is one of the most fascinating things i've seen ai do. 🤯 in 1809, someone sent a secret coded letter to one of napoleon's generals 24 rows of numbers, symbols, and letters that no one could read. it stayed unsolved for 217 years because cryptographers knew about it but never had the time to sit down and do the work. last week, a guy uploaded a single photo of the letter to gpt-6 astra and just said: solve this. in 6 hours, astra read 1,300 handwritten symbols from a 217-year-old document, found a key that no one had ever linked to this letter, wrote its own code to crack the remaining cipher, and translated the full message from french. the letter turned out to be a war briefing sent two weeks before austria invaded. and it finished a sentence in napoleon's own memoir that had been incomplete for over 150 years historians had the first half, this letter had the rest. a leading cipher historian verified the solution and officially marked it as solved. the person who cracked it wasn't a cryptographer. he was just curious and had the right tool.
This post quotes another post: “GPT-6 Astra decrypts a 217-year-old cipher to one of Napoleon's generals”