In London’s Imperial War Museum, a team staged a curious experiment: they fed a modern AI the problem that haunted Alan Turing during the darkest days of World War II and got a result in the time it takes to have a cup of coffee.
This is a reminder of how far computing has come. But it’s also a big question mark for where encryption goes from here on.
Apparently Less Enigmatic Nowadays
The Enigma machine, with its three rotating rotors and a labyrinthine plugboard, was Nazi Germany’s pride. It was considered to be one of their symbols of invulnerability. When German generals tapped out encrypted messages each morning, they trusted that the ever-changing complexity of their code—different settings each day, no letter ever mapping to itself—would be more than the Allies could solve.
That confidence wasn’t absurd. The mathematics behind Enigma was daunting. Each rotor could be set to one of 26 positions, and combinations of plugboard wiring created over 150 quintillion possible configurations.
“The number of possible ways in which a message could be encrypted was astronomically large,” Prof. Michael Wooldridge, a computer science expert at the University of Oxford told The Guardian. “Far, far too large for a human to exhaustively check.”
Cracking the Enigma was a mammoth challenge. That’s why it took Turing and his colleagues at Bletchley Park months to gain a foothold. After the groundwork laid by Polish mathematicians, the British team built their own machines—nicknamed “bombes”—which automated the brute-force search for valid Enigma settings. At peak efficiency, they could decrypt about two messages per minute. But every second mattered in times of war, and the cracking of the Enigma code was a pivotal moment.
“To be able to crack it—it took them months, more than a year—but to be able actually to do this within the lifetime of the war, it was a huge thing,” said Dr. Mustafa A Mustafa, a senior lecturer in software security at the University of Manchester. “God knows what would have happened if we hadn’t cracked Enigma in time.”
Modern Machines, Ancient Ciphers
Fast-forward 80 years. What took bombes hours, modern processors dispatch in moments. One AI model, trained on German using Grimm’s Fairy Tales and aided by 2,000 cloud servers, decrypted a four-rotor Enigma message in just under 13 minutes.
<!– Tag ID: zmescience_300x250_InContent_3
–>
Unlike the rigid brute force methods of Turing’s day, this AI used probabilistic reasoning—comparing decrypted outputs to patterns in the German language. “The AI was trained to look for German language, and then work out the statistical probability of the sentence decrypted being the accurate original,” Lukasz Kuncewicz, Head Data Scientist at Enigma Pattern, the company behind the demonstration, told Tech Radar.
Wooldridge added that the whole process could be replicated even more efficiently using contemporary computing models. “It would be straightforward to recreate the logic of bombes in a conventional program,” he said. “Then with the speed of modern computers, the laborious work of the bombes would be done in very short order.”
That doesn’t mean Turing’s legacy is diminished. Far from it. His pioneering work cracked a wartime code and laid the foundation for the very computers that render it obsolete today. He is, after all, considered the father of artificial intelligence.
“The bombes were crude hardwired mechanical computers,” Wooldridge noted. “And the power of modern datacentres is hard to imagine. Enigma would not remotely be a match for these.”
The problem is how our modern encryption stands up to potential attacks.
What does this mean for encryption?
While cracking of Enigma might seem like a historical footnote, it raises pressing questions for today’s digital age.
If AI can unravel what was once unbreakable, what codes remain secure?
For better or for worse, encryption systems like RSA keep the internet secure for now. But RSA was developed in the 1970s and based factoring large prime numbers, so it’s hard to tell how long it will still hold — especially with quantum computing just around the corner!
“In the case of RSA, it’s the problem of factoring very large numbers. Brute force techniques—looking through all the alternatives—just won’t work on these problems,” Wooldridge said. “If quantum computers ever deliver their theoretical promise, then we may need completely new techniques to keep our data safe.”
This is no longer the domain of spies and generals. As AI becomes more capable, the same techniques used to crack Enigma could potentially be turned on passwords, personal data, and secure communications. “It’s like a knife,” Kuncewicz said. “It can save lives or it can take lives.”
For now, most AIs require human oversight. The AI used in the Enigma test still needed engineers to check that it hadn’t gone off-track—declaring nonsense as meaningful. But this boundary is becoming increasingly thin.
Would Turing be amazed? Perhaps. But he might also have seen it coming.
