ChatGPT-6 Astra Cracks 108-Year-Old WWI German Code for the First Time

9 min read

A 108-year-old German radio message from the final days of World War I has reportedly been deciphered with the help of ChatGPT-6 Astra, providing a new interpretation of a coded military transmission that had remained unresolved for decades.

The message was encoded using the German military’s ADFGVX cipher, a complex system that was used during World War I to protect military communications. According to the developer who conducted the analysis, ChatGPT-6 Astra identified a previously overlooked encryption key and produced a German plaintext referring to the arrival of an English cruiser at Sevastopol and the expected arrival of an Allied squadron.

The proposed decoding was then compared with historical military records. Those records reportedly show that the British cruiser HMS Canterbury arrived at Sevastopol in Crimea on November 24, 1918, while an Allied squadron arrived two days later.

The result has attracted attention because it illustrates how modern AI systems can combine historical information, cryptographic techniques and pattern recognition to revisit old problems.

What Was the 108-Year-Old German Code?

The encrypted message was a German military radio transmission dating from November 1918.

It was encoded using the ADFGVX cipher, a system named after the six letters used in its encryption grid: A, D, F, G, V and X.

The cipher was introduced by the German military during World War I and was designed to make intercepted radio communications difficult for Allied intelligence services to read.

ADFGVX combined a fractionation system with a transposition process. The system could turn ordinary letters into pairs of the six designated characters, after which the resulting sequence was rearranged using a keyword.

This made the cipher substantially more complicated than a simple substitution cipher.

Although many German military messages using the system have been decoded, a number of individual messages have remained unresolved.

The message examined by the AI developer was reportedly one of those outstanding examples.

How ChatGPT-6 Astra Decoded the Message

The reported breakthrough involved identifying the encryption keyword “TRUPPENVERSCHIEBUNG.”

The German word roughly relates to troop movement or troop transfer.

According to the published account, the key had already been documented as a German military cipher keyword. However, researchers had apparently assumed that it was not used until a later date.

The encrypted message itself was transmitted before that assumed starting date.

That assumption may have prevented previous attempts from testing the keyword against this particular message.

ChatGPT-6 Astra reportedly challenged that assumption and used the keyword to reconstruct the cipher table and reverse the encryption process.

Once the key was applied, the resulting German text appeared to describe naval movements around Sevastopol.

What Did the Decoded Message Say?

The proposed plaintext was:

“EIN ENGLISCHER KREUZER EINLIEG X SEWASTOPOL X S4STEN X EIN GESCHWADER DER X ALLIIERTEN FOLGT 26STEN X.”

The reported English interpretation is approximately:

“AN ENGLISH CRUISER ARRIVED AT SEVASTOPOL ON THE ?4TH. AN ALLIED SQUADRON FOLLOWS ON THE 26TH.”

One character in the date remains uncertain in the proposed reading.

The uncertainty is significant because the decoded text appears to contain an unusual character where a number would normally be expected.

However, the historical information surrounding the message provided an opportunity to test the proposed interpretation.

HMS Canterbury Provides a Historical Check

One of the most interesting parts of the reported discovery was the comparison with historical naval records.

The decoded message appeared to indicate that an English cruiser arrived at Sevastopol on November 24, 1918.

Historical records reportedly show that HMS Canterbury arrived at Sevastopol on November 24.

The decoded message also referred to an Allied squadron arriving on November 26.

That date reportedly corresponds with historical records of an Allied squadron arriving in the area.

The correspondence between the decoded message and the historical timeline provides an important piece of supporting evidence for the proposed solution.

It does not, by itself, constitute absolute proof that every character in the message has been reconstructed correctly. However, the matching dates make the proposed interpretation considerably more interesting than an arbitrary plaintext that merely fits the cipher mathematically.

Why Was the Message Not Decoded Earlier?

The reported explanation involves an assumption about when the encryption key was used.

Historical research had associated the keyword TRUPPENVERSCHIEBUNG with German military communications beginning around December 9, 1918.

The message under investigation, however, was transmitted earlier in November.

If researchers assumed that the keyword could not have been used before December 9, they would have had little reason to test it against a message from November.

That creates an intriguing example of how historical assumptions can affect cryptographic research.

A cipher can remain unresolved not necessarily because the underlying mathematics is impossible, but because researchers may be searching within the wrong set of assumptions.

The AI system reportedly had fewer reasons to respect that historical assumption and therefore tested a possibility that human researchers may have overlooked.

What Is the ADFGVX Cipher?

The ADFGVX cipher was one of the more sophisticated field ciphers used by Germany during World War I.

Its name comes from the six letters used by the encryption system.

A message would first be transformed using a Polybius-style square. Each plaintext character would correspond to a pair of ADFGVX symbols.

The resulting sequence was then subjected to a transposition process using a keyword.

This two-stage approach made the encrypted message difficult to decipher without knowledge of the correct procedure and key.

The system was particularly useful for military radio communications because radio messages could be intercepted by enemy forces.

The Allies eventually became highly effective at breaking German military communications, but individual encrypted messages could still present challenges depending on the available information.

Why the AI Discovery Matters

The significance of the reported discovery is not simply that an AI system decoded an old message.

The more interesting aspect is the method used to reach the proposed solution.

AI systems can process large quantities of historical material and rapidly test different hypotheses. They can also connect information that might normally be separated across cryptographic references, historical records and military documents.

In this case, the system reportedly identified a potential cipher key, applied it to the encrypted message and then checked whether the resulting plaintext made historical sense.

That combination of cryptographic reasoning and historical cross-checking is potentially useful for researchers examining unresolved historical documents.

The Importance of Independent Verification

Despite the excitement surrounding the result, independent verification remains important.

The decoding was reported by the developer who performed the experiment, and the available reporting describes it as a proposed solution rather than a universally accepted historical fact.

A convincing cryptographic solution should ideally be reproducible by independent researchers.

Other cryptographers can test the same ciphertext, apply the proposed key and determine whether the resulting plaintext follows consistently from the known ADFGVX procedure.

Historians can separately examine the underlying military records to determine whether the dates, locations and movements correspond to the decoded information.

If independent researchers reproduce the result, confidence in the solution would increase substantially.

AI and Historical Cryptography

The case demonstrates a potentially important application of artificial intelligence in historical research.

Many archives contain encrypted documents, damaged manuscripts and incomplete records that have received relatively little modern analysis.

Traditional cryptanalysis can require substantial expertise and time.

AI systems can potentially assist by generating hypotheses, searching historical references and testing multiple interpretations.

However, AI-generated solutions also require careful verification.

A language model can produce plausible-looking text even when a proposed interpretation is incorrect. For historical cryptography, therefore, the strongest results are those that can be independently reproduced and matched against external evidence.

The reported HMS Canterbury connection is particularly relevant for this reason.

A Message From the End of World War I

The timing of the message adds another layer of historical interest.

The transmission came in November 1918, shortly after the Armistice that ended the fighting on the Western Front.

The political and military situation around Crimea and the Black Sea remained complicated, however, with Allied forces involved in the region during the postwar period.

A German naval intelligence message tracking Allied movements around Sevastopol therefore provides a snapshot of military communications during a turbulent period of European history.

The decoded information appears to concern movements that were occurring in real time.

More than a century later, the same information can now be compared with historical records to evaluate the accuracy of the proposed reading.

What the Discovery Does Not Mean

The reported breakthrough should not be interpreted as evidence that modern AI can automatically break all historical or modern military encryption.

The ADFGVX cipher is fundamentally different from modern cryptographic systems.

The reported solution also appears to have benefited from existing historical knowledge about German cipher keys.

In other words, the achievement involved more than simply asking an AI to “break” an unknown cipher from scratch.

The system reportedly connected an existing cryptographic key with a message that researchers had not previously decoded using that key.

That distinction is important when considering what the experiment says about modern AI capabilities.

Conclusion

The reported ChatGPT-6 Astra 108-year-old WWI code breakthrough offers a fascinating example of AI-assisted historical cryptanalysis.

A German radio message encoded with the ADFGVX cipher has reportedly been deciphered using the keyword TRUPPENVERSCHIEBUNG, revealing information about an English cruiser arriving at Sevastopol and an Allied squadron expected to follow.

The proposed interpretation is particularly notable because historical records reportedly match the decoded dates, including the arrival of HMS Canterbury at Sevastopol on November 24, 1918.

The discovery also suggests that an earlier assumption about when the encryption key was used may have contributed to the message remaining unresolved.

Whether the solution ultimately becomes accepted by independent cryptographers and historians will depend on further verification. Nevertheless, the case demonstrates how modern AI can be used to revisit historical puzzles that have resisted previous attempts.

More broadly, the episode highlights a growing role for AI in historical research: not simply generating answers, but testing overlooked possibilities, connecting fragmented evidence and helping researchers examine documents from the past with new tools.

Leave a Comment