OpenAI’s GPT-6 Astra has decoded a 217-year-old secret letter written by Napoleon’s stepson, and the AI’s solution checks out against known historical records. The breakthrough gives historians a practical way to read coded messages from the past, and it carries the weight of a Rosetta Stone for modern cryptography.
The Letter From Eugène to Marmont
The letter was sent in 1809 on Napoleon’s request by his stepson Eugène, the viceroy of Italy. It was addressed to Marshal Auguste de Marmont, who was stationed far from any other French troops on the eastern shore of the Adriatic Sea. The message was meant to tell Marmont the position of every French and allied army across the continent.
To keep the enemy from learning those details, the letter was written in code. The message was intended to be read only by Marmont, not anyone else who might intercept it.
Carter Church Cracks the Code
Carter Church, an AI engineer at cybersecurity firm Sentinel One, found the letter while looking for a code to break in order to show off GPT-6 Astra’s abilities. He gave the model one prompt and one image of the letter, and six hours later, the task was done.
“I didn’t believe it at first,” Church told Live Science in a message. “Everyone has the experience of seeing a model claim something outrageous, but the difference here though is that the solution is mechanically testable: there’s essentially zero chance an incorrect key could produce coherent and historically consistent output.”
The Code Was Hard to Read
The letter contained 155 different signs, making up 1,300 signs in total. The code is homophonic, meaning common letters have several interchangeable symbols. That makes it hard to guess what each sign stands for without a key.
The only publicly available image of the letter was a low-resolution scan published in a French army journal in 1969. Church said that low-resolution scan and a single prompt were all GPT-6 Astra needed to transcribe the text and find a partial table by cryptology historian Daniel Tant.
Astra Finds the Partial Table
Astra made the connection to Tant’s table itself via web research as it was investigating. Tant had catalogued the table as “Code entre Eugène de Beauharnais et le Général Grammont,” which may have been a naming mistake — Grammont versus Marmont. Either way, Astra identified this example and tried applying it to the cipher, and it worked.
The partial table documented what 33 of the symbols meant, accounting for about a third of the letter. That was enough for Astra to determine what the other symbols stood for and translate the message.
Michael Rowe on the Rosetta Stone Comparison
Michael Rowe, a researcher in European history at King’s College London, called the findings “impressive.” He noted that because we have access to Napoleon’s original correspondence to his stepson, where he detailed what the ciphered message should include, we can be fairly confident in the translation of Eugène’s letter.
“It’s a bit like a kind of Rosetta Stone,” Rowe told Live Science. “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. Then if you use the exact same kind of tool and methodology on something where you don’t know [the contents], you’re more likely to trust the outcome.”
Rowe also pointed out that the translation isn’t too surprising to historians. “The cipher version just repeated a large segment of what’s in that [original] letter, so in a sense it doesn’t tell us anything new — it’s not like it reveals some kind of great new information,” Rowe said.
What the Letter Actually Said
The translated letter encouraged Marmont not to fear opposition. “When you receive the order to march, you must not allow yourself to be intimidated by a few troops or a gathering of rabble,” the newly-translated letter reads.
The message also included the positions of troops and their commanders, giving Marmont a full picture of where the rest of the French armies stood.
Zack White on Napoleon’s Mindset
Zack White, a history research fellow at the University of Portsmouth and host of the Napoleonic Wars Podcast, said the translation raises “some curious questions about Napoleon’s mindset in the run up to the campaign in Austria in 1809.” For example, Napoleon actually had 60,000 fewer troops available to fight at the time than is claimed in this letter, and he also seems fixated on the number of troops Austria might send to invade Italy.
“It’s important that historians now set the AI’s works in a bigger picture of the history of the period,” White said, adding “it will also be interesting to see if the key that has resulted from this AI’s work can now be used on other correspondence.”
The Limits of the Breakthrough
Church said the hardest part wasn’t generating potential solutions — it was verifying them. “The verifiability of solutions is actually the most difficult part, far more difficult than generating potential solutions,” Church said. “You simply cannot automate accountability.”
That caution matters. A model can propose a translation, but a human still has to check it against known history before accepting it as correct.
Why This Matters
Ciphers were common during this period, used to convey sensitive military and diplomatic information. “French Emperor Napoleon Bonaparte was forced to use ciphers as a way of communicating sensitive information securely across his vast empire,” White said. “It was always a very real possibility that these letters would be intercepted, potentially providing vital intelligence to France’s enemies about the Emperor’s plans.”
Using AI to crack these codes could offer insights into the private lives of historical figures, White said. Ciphers were often also used for private correspondence, not just official orders.
The Future of Decoding History
The success of this project shows what AI can do when applied to historical puzzles. Astra’s ability to connect a fragmentary table to a specific letter and then apply it correctly suggests the technique could work elsewhere.
Historians will now test the method on other coded documents. The key produced from this work could unlock correspondence that has remained unreadable for decades.
The letter itself isn’t a surprise to scholars. It repeats information already known from Napoleon’s other letters. But the method of reading it is new, and it proves that AI can serve as a modern Rosetta Stone — a tool that lets us understand texts we couldn’t otherwise decipher.
Key Facts Box
- Letter sent: 1809
- Author: Eugène, viceroy of Italy (Napoleon’s stepson)
- Recipient: Marshal Auguste de Marmont
- Code type: Homophonic (common letters have multiple symbols)
- Symbols in letter: 1,300 total, 155 distinct
- Partial table found: Documented 33 symbols, about a third of the letter
- Time to decode: Six hours
- Model used: GPT-6 Astra
The breakthrough is a reminder that history hides in plain sight. A letter scanned once in 1969 took over two centuries to read, and it took an AI to finish the job.
Source material: “'A kind of Rosetta Stone': AI decodes 217-year-old secret letter ordered by Napoleon,” Live Science.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.