Back to all articles
•Technology

AI Models Crack Enigma Code Left Unsolved Since 2005

View original source

In a remarkable testament to modern technology's crossover with historical intrigue, AI models from OpenAI and Anthropic have successfully decoded messages from Nazi Germany's Enigma machines, longstanding unsolved since World War II.

  • Alan Turing, renowned for his role in cryptanalysis during WWII, laid the groundwork with the Bombe, an early computer that decrypted many Enigma messages. A few, however, remained unsolved due to errors by encoders.
  • Recently, Carter Leffen, an AI developer, enlisted OpenAI's AI model, Astra, to identify and decode these messages. Astra managed to solve a particularly troubling message first baffled by researchers in 2005. Link to discovery.
  • Frode Weierud, an expert with a dedicated cryptology website "Crypto Cellar", validated Leffen's breakthrough, expressing astonishment at the accuracy and speed of Astra. His validation highlighted Astra's in-depth archival research capabilities.
  • Anthropic's Claude Opus 5, used by cryptanalyst Jack Willis, solved another Enigma message by leveraging known officer name signatures.
  • Weierud remarks on the potential nearing end of unsolved Enigma messages, with only seven remaining.

With AI acting as formidable cryptanalysts, these developments signify a leap in both historical understanding and technological prowess.