GPT-6 Astra Decodes 1941 Enigma Radio Message

In Crypto Regulations
September 19, 2026

GPT-6 Astra Decodes 1941 Enigma Radio Message

Bloomberg’s Carter Leffen reported that using GPT-6 Astra, he reconstructed the text of a German military radio message from July 10, 1941. The message had remained undeciphered in the CryptoCellar archive since 2005.

The radio message, identified as MVUEH, contains 82 encrypted characters. Two known keys used on the same day did not work for it.

The research took place on September 14-15. Leffen set the task and guided the work, while Astra and several parallel AI agents studied archival materials, compared interpretations of poorly legible characters, wrote programs to find the key, and verified the solutions.

Leffen estimated Astra’s direct work at about 10 hours. However, in a detailed report, he noted that processes were run in parallel, making it impossible to precisely separate the model’s work from human input.

A Place Name as a Clue

A key hint came from another radio message from July 10, 1941—SIPVX, which was deciphered in 2017. The name Rosenow appeared twice consecutively in its text.

Researchers hypothesized that the sequence ROSENOWROSENOW might also be present in MVUEH. These 14 characters were used as a “crib”—a presumed fragment of the original text that helps reduce the number of possible keys.

The remaining 68 characters of the message were not known in advance.

The condition of the archival documents complicated the work. Before searching for the key, the team identified 12 places where individual handwritten letters could be read differently. The final reconstruction differs from one of the previous transcriptions by three characters.

Astra wrote an Enigma simulator and cryptanalysis programs in Python and C++. Under the given conditions, the system checked about 4.29 billion combinations of crib positions and rotor parameters.

Interactive Enigma model with rotors II–V–III and settings used for decrypting the MVUEH 1941 radio message
Interactive visualization of Enigma with restored settings for the MVUEH 1941 radio message. Source: MVUEH.

After identifying a candidate, about 14.8 million physical keys were separately checked against the preserved radio message header.

As a result, researchers restored the rotor order II-V-III, ring settings H-M-F, initial position R-W-D, and ten plugboard connections.

Two independent Enigma implementations were able to decrypt and re-encrypt all 82 characters of the message and its header with the found parameters.

Contents of the Message

In the radio message, the sender requested information on the further route, indicated being in Rosenow, and demanded an immediate radio response.

The specific location referred to by the name is not yet determined. Researchers tentatively read the sender’s signature as Waschbusch.

The result was verified by Enigma researcher Frode Weierud, who manages the CryptoCellar archive. He confirmed the found key and noted that it significantly differs from the two other known keys for July 10, 1941.

According to him, previous attempts might have been hindered by errors in the old transcription of the radio message and an unusual left rotor transition near the end of the message.

The experiment’s authors emphasize that Astra did not “break Enigma” as a cipher. The cryptanalysis methods used were developed before and during World War II. In this case, researchers restored the key and text of a specific radio message that remained unsolved in the modern archive.

They also do not claim the first-ever decryption of such a message: the search was complete only within the assumptions about the machine type, rotors, plugboard, reading options of the original document, and the chosen crib.

Earlier in September, OpenAI published a claimed solution to the existence and smoothness of the Navier-Stokes equations—one of the seven Millennium Prize Problems, each with a $1 million prize set by the Clay Institute.

Avatar photo
/ Published posts: 1111

Steven M. Crimmins is a cryptocurrency strategist and freelance writer who has followed the blockchain industry since Bitcoin’s early days. Known for his sharp analysis of altcoins and trading strategies, Steven provides Satoshi News Africa readers with market-focused content grounded in research. He is especially interested in how African traders are adopting crypto as an alternative to traditional markets. Steven is also a podcast host, where he discusses emerging technologies and investment trends.