AI Models like GPT-6 Astra Crack 433-Year-Old Papal Secret Letters, Unveiling Royal Family Insights
1 day ago / Read about 0 minute
Author:小编   

GPT-6 Astra and Claude Opus 5.5, two advanced AI models, independently deciphered two highly confidential letters from 1593, penned by the Catholic League, almost at the same time. These letters had long remained a mystery, unsolved for centuries. The encryption technique used was a homophonic substitution cipher, where each symbol represented two letters, posing a complex challenge with four distinct layers: deciphering handwritten symbols, mapping digraphs (pairs of letters), restoring Old French language, and reasoning within the historical context.

To tackle these challenges, the two AI systems meticulously constructed annotated datasets by drawing on historical decryption records. They also refined pronoun references by integrating historical context, ultimately achieving near-complete restoration of the letters' content. The decryption process shed light on the political maneuvering behind King Henri IV of France's conversion to Catholicism and his subsequent rise to the throne. It was revealed that a political deal involving 500,000 écus was struck, intertwining the interests of the Pope, the Spanish royal family, the French Catholic League, and exposing political betrayals at the heart of Roman power.

Previously, AI had also demonstrated its prowess by swiftly cracking Napoleon's lost coded battle reports from 1809. Given that global archives are brimming with countless unsolved encrypted collections and human decryption efforts have been inefficient, AI is emerging as a super decryption tool for historians. It is poised to uncover hidden historical data on a large scale, prompting a re-examination and retelling of established historical narratives.