Two independent researchers have demonstrated a new breakthrough in using artificial intelligence for historical research. Cryptoanalysts use their latest LL.M. to solve problems that have eluded human experts for years.
OpenAI’s Astra model was able to independently detect archival clues and reconstruct the text of messages that had been illegible since 2005, reducing weeks of manual work to just two days.
In turn, cybersecurity expert Jack Willis applied the Claude Opus 5 model, using officials’ personal signatures to guide their efforts. Famous cryptography researcher Frode Weyerud confirmed the correctness of the results and stated that after these successes, there are only 7 encrypted Enigma messages left in the world archives that have not been cracked.
