GPT-6 Astra Decodes 217-Year-Old Napoleonic Code in Six Hours, Reveals Lost Military Orders

AI Decodes Napoleonic-Era Cipher That Stumped Historians for Two Centuries

AI has unlocked a military secret hidden for more than 200 years. Carter Church, an engineer at cybersecurity firm SentinelOne, used OpenAI's GPT-6 Astra model to decipher a 217-year-old Napoleonic code in approximately six hours.1

1

The encrypted letter, sent in March 1809 from Eugène de Beauharnais, Napoleon's stepson and Viceroy of Italy, to French General Auguste de Marmont, had sat unsolved on Cryptiana's Unsolved Historical Ciphers list for decades.1

1

What makes this achievement remarkable isn't just the codebreaking itself, but that GPT-6 Astra completed the entire multi-modal workflow from a single image and prompt.5

5

How GPT-6 Astra Cracked the Homophonic Cipher

The document contained 24 rows of numbers, letters and invented symbols—1,300 cipher units spanning 155 unique symbols in total.4

4

The only publicly available source was a low-resolution scan published in a French army journal in 1969.2

2

The homophonic cipher made conventional frequency analysis unreliable because common letters had several interchangeable symbols, and some signs represented entire words.4

4

Astra began by transcribing handwritten symbols from the scanned image, cutting the 1,202 by 1,836 plate into rows and determining which of 175 hand-drawn marks were the same sign written twice.3

3

Through web research, the AI identified a partial cipher key catalogued by cryptography historian Daniel Tant that documented 33 of the symbols, accounting for about a third of the letter.2

2

Using multi-modal processing, Astra then cracked the rest with its own simulated-annealing solver, developing a computational cryptanalysis program and applying statistical regularities drawn from French literature to recover the remaining characters.1

1

4

4

What the Decoded Letter Reveals About Napoleon's Military Strategy

The deciphered message reveals troop positions and military secrets from a pivotal moment in European history. In March 1809, Austria was preparing to fight France while a large share of French forces remained tied down in Spain.3

3

Marmont, commanding French troops in Dalmatia on the far side of the Adriatic, risked being isolated if Austria attacked.3

3

On March 16, Napoleon ordered Eugène to inform Marmont of the positions of main French and allied forces, specifying the message be written in cipher and carried by a reliable officer.3

3

The decoded text reads: "The Russians are marching against Austria, and the army [of] Friuli today numbers eighty thousand men. Add the Dalmatian corps, without counting the reserves being organized everywhere in the rear, and you see that Austria will be attacked at once on every side."3

3

The letter revealed reported positions of forces commanded by Davout, Masséna, and Oudinot, along with claims about Russian military movements.4

4

It also urged Marmont not to be "intimidated by a few troops or a gathering of rabble" when receiving orders to march.3

3

Historical Significance and Verification of AI's Work

The freshly deciphered text fills in missing context from Napoleon's 1865 memoir, which includes part of Napoleon's order relating to preparation for the Austrian invasion but breaks off mid-sentence saying Marmont mustn't be held back by "a handful of ..."1

1

The deciphered letter completes this sentence with "a gathering of rabble."1

1

Satoshi Tomokiyo, who maintains the Cryptiana historical-cipher database, reviewed the solution and marked the entry as solved, though five symbol interpretations remain uncertain.4

4

Church published his complete research record on September 18, including transcription, cipher tables and scripts that allow others to reproduce the result.3

3

"The solution is mechanically testable against the ciphertext," Church explained, noting there's essentially zero chance an incorrect key could produce coherent and historically consistent output.2

2

Why This Matters: AI as a Modern-Day Rosetta Stone

Historians are calling this achievement a modern-day Rosetta Stone for translating coded historical messages. Michael Rowe, a researcher in European history at King's College London, noted that because we have access to Napoleon's original correspondence to his stepson detailing what the ciphered message should include, we can be fairly confident in the translation.2

2

"It's a bit like a kind of Rosetta Stone," Rowe explained. "If you've tested out AI on this letter, where you pretty much know what the contents would have been because it's referred to in another letter, the AI does its thing and essentially reproduces what you're expecting to see."2

2

The single-prompt AI approach demonstrates how artificial intelligence can combine image recognition, research, programming and verification in one workflow, suggesting how it could help scholars unlock other neglected historical archives.4

4

Church observed that while the cipher itself isn't difficult and cryptographers could have cracked it given time and energy, the work to transcribe, catalog, translate and then crack is involved and multi-disciplinary.5

5

"The takeaway here is that the people shortening those lists will increasingly be outsiders who were simply curious," Church noted.5

5

Watch how AI continues to tackle complex historical ciphers and what other unsolved mysteries might be cracked next using similar multi-modal approaches.