A Vatican manuscript unread for 400 years was decoded by an AI tool in just 29 minutes
A team of international researchers used AI to decode the Vatican manuscript as part of the Descrypt project
For more than 400 years, a 408-page manuscript sat untouched in the Vatican library, its pages filled with 34 cryptic symbols that no scholar had been able to decipher. An AI tool solved it in just 29 minutes.
A compendium of curious cures
The document, known as the Borg cypher — named after the Vatican's Borg collection — turned out to be a compendium of unusual medical treatments.
Among its prescriptions were recommendations to drink several glasses of red wine or ferment nutmeg in dough as remedies for dysentery.
Beáta Megyesi, Professor of Computational Linguistics at Stockholm University, was part of the team that decoded the manuscript.
She describes the work as "detective work where every symbol, pattern, and partial solution may bring us closer to someone's secrets and to a lost historical world."
A vast trove of encrypted history
The Borg cypher is far from an isolated case. Roughly one per cent of all material held in libraries and archives worldwide is fully or partially encrypted — a figure that represents thousands of documents that have never been read in the modern era.
The scale of the challenge is illustrated by the experience of Cécile Pierrot, a cryptologist at INRIA in France, who spent six months decoding a three-page letter written by Holy Roman Emperor Charles V using 120 different cypher symbols.
The reward for her team's painstaking work was the discovery that one of the most powerful rulers in history had been living in terror of an assassination plot by a French-allied Italian mercenary.
Even a straightforward two-page transcription of unfamiliar symbols takes Pierrot an entire day to complete.
How the AI system works
According to Megyesi and colleagues working on the multinational Descrypt project, the AI technology being developed for this field attacks the problem in two ways simultaneously.
Platforms such as Transkribus can analyse handwritten documents by detecting text blocks and lines and transforming symbols into machine-readable text.
The deciphering process then involves applying several algorithms to calculate the frequencies and distributions of certain symbols and establish what type of cypher has been used.
The most advanced system developed by the researchers combines AI chatbot technology with decryption algorithms that draw on cypher and plaintext examples, large language models trained on historical texts, and image recognition algorithms capable of processing handwritten annotations.
Crucially, the technology is also able to self-improve by incorporating expert corrections following each analysis.
When tested on the Borg cypher, the system successfully analysed, translated, and explained a 500-symbol extract within 30 minutes.
A track record of breakthroughs
The same methods had previously been employed to decrypt a 17th-century letter from the Thirty Years' War and an 18th-century document relating to the rituals of a German secret society.
The team also assembled a collection of 400 coded postcards dating from the late 19th century, several of which were subsequently deciphered as German love letters.
Mysteries that remain unsolved
Not every ancient cypher has yielded to modern methods. The Phaistos Disc, a clay artefact dating back 4,000 years from the island of Crete, continues to resist all attempts at decipherment.
The ancient Greek writing system known as Linear B has also yet to be fully unlocked. By contrast, coded letters written by Mary, Queen of Scots — decoded only recently — confirmed that she had been actively involved in conspiracies to reclaim the throne and shed new light on her fraught relationship with her son, James I of England.