Claude Opus, GPT‑6 Astra crack unsolved Enigma messages
Two independent researchers used the models to break World War II-era Enigma ciphers that had defeated cryptanalysts for decades, according to TechCrunch, which said seven messages remain unsolved.
Two cryptanalysts used AI models to break previously unsolved Enigma-encrypted messages from World War II, TechCrunch reported Friday. Carter Leffen used OpenAI's GPT-6 Astra to crack a message that had "baffled researchers since 2005." Jack Willis used Claude Opus 5 on September 21 to break a separate message, using an officer's name as a key.
Astra conducted archival research, identified context clues and built an Enigma simulator to recover the plaintext, according to TechCrunch. Frode Weierud, a retired engineer, maintains the Crypto Cellar database of unsolved Enigma traffic. He said GPT-6 Astra was "behaving like a very professional cryptanalyst and archive researcher."
Weierud told TechCrunch he had spent "several weeks" researching archival files that Astra worked through in two days. TechCrunch said seven Enigma messages remain unsolved. One additional message has known plaintext, but its underlying code has not been recovered.
The messages are part of a long-running volunteer effort to recover unbroken wartime Enigma traffic. Neither OpenAI nor Anthropic commented on the results. TechCrunch said the results were checked against Weierud's own archival research.