Quantum Computers Are Poised to Crack Encryption. AI Could Beat Them to It.

What if the mathematical locks and keys that secure our bank accounts, email inboxes and everything else on the internet turn out to be a lot less secure than we thought? What if the password-protected internet we rely on every day collapses?

Quantum computers have long threatened such an apocalypse. For decades, engineers have been preparing for their arrival—and their inevitable abuse by well-heeled bad guys.

The result is known as “postquantum cryptography.” According to leading mathematics experts, an orderly transition of the world’s digital systems to this new paradigm should avert an event that could otherwise send us back a century, technology-wise.

But now there is a potentially more pressing issue: AI’s impressive math skills.

Artificial intelligence is pushing advances in various mathematics fields at an unprecedented rate, conquering problems that our puny human minds have wrestled with for centuries. The fear is that this could include new solutions to the math problems that keep our encrypted communications encrypted.

For encryption to continue to work, our assumptions about the algorithms that make it possible must remain true. One such assumption is that some mathematical operations are easy to solve in one direction, yet hard to reverse. The classic example: Multiplying two large prime numbers is simple enough, while factoring the very lengthy result back into those two prime numbers is essentially impossible with conventional computers.

But some cryptographers argue that we might not be able to count on this remaining the case.

“A lot of what makes us feel like cryptography is secure is because a lot of smart people have looked at crypto for a long time, and we didn’t find anything,” says Matthew Green, a cryptographer and associate professor at Johns Hopkins University, on mathematicians’ ongoing search for code-breaking formulas. “But what we’re learning is that human beings looking at things is not the gold standard—it’s not relevant anymore.”

It’s reminiscent of the cryptographic race during World War II. Alan Turing and his team at Bletchley Park, innovating on an electromechanical system pioneered by Polish engineers, found a way to crack Germany’s infamous Enigma code. Many great minds thought Enigma was unbeatable, but Turing’s device proved them wrong.

Green cites recent breakthroughs in mathematics by AI labs, including a partial solution to the infamously challenging Navier Stokes equations, and a more recent breakthrough disproving the Exponential Time Hypothesis.

It’s this second result that has some researchers especially worried. This long-held hypothesis suggests that certain problems don’t have shortcuts. The only way to solve them is to take the long way. But if the hypothesis is false, it could render decades of research in theoretical computer science irrelevant. If there are shortcuts to solving these problems, then those shortcuts could make it far easier to crack “uncrackable” encryption.

Vitalik Buterin is the creator of the Ethereum protocol which underpins much of modern cryptocurrency systems. In early October, before the latest breakthrough math results from OpenAI and Anthropic, he issued a warning on X: “We should take the risks to cryptography from AI-accelerated math seriously, and minimize our exposure to not just quantum-vulnerable cryptography, but also potentially AI-vulnerable cryptography.”

On Monday, he reiterated the warning at the Ethereum Foundation’s annual conference in Mumbai, calling for “post-AI cryptography.”

Green cautions that there’s no evidence the world’s cryptographic algorithms have already been cracked, but he thinks the next AI-powered math breakthroughs could touch more directly on cryptography’s foundational math.

How worried should we be? That’s the subject of fierce debate among experts.

Baseless speculation about an impending collapse of the world’s online security isn’t just premature, it can be irresponsible, argues Yehuda Lindell, head of cryptography at Coinbase and a computer-science professor currently on leave from Israel’s Bar-Ilan University.

“Anything is possible,” says Lindell. “But speculating that bad things can happen is something which just doesn’t help anyone.”

Lindell objects in particular to claims that postquantum cryptographic algorithms, which are gradually being implemented, might be vulnerable enough to AI’s mathematical discoveries to force us to rethink the safeguards. He says evidence points to the opposite: These algorithms are likely to withstand most attacks by any kind of computer. Also, he says it is possible AI will help us strengthen our cryptographic algorithms.

At worst, he adds, engineers might have to lengthen cryptographic keys, the numbers used in secure exchanges. This would slow down the encoding and decoding that constantly occurs in our online interactions, but it wouldn’t make things unusable.

The reason cryptography experts find themselves at odds is because of a disagreement about fundamental mathematical questions that remain unresolved.

Those like Green, who are concerned over cryptographic algorithms collapsing under AI innovation, worry that somewhere out there in the world of mathematics is an as-yet-undiscovered white whale. If a discovery is made that makes even some “impossible” math problems solvable with today’s computers, our digital locks could weaken.

Experts on Lindell’s side believe decades of work by mathematicians suggest such a possibility is vanishingly small. “My claim isn’t that AI will not help with cryptographic analysis—I think it very likely will,” he says. Yet he believes the fundamental mathematical assumption that holds up his entire field—that there really are hard problems that will never be solved by a shortcut—will stand the test of time.

Even in the worst-case scenario, Lindell would rather an AI lab like OpenAI or Anthropic make the discovery that modern cryptography is insecure, since these companies have a history of disclosing vulnerabilities so that engineers can patch them.

“We don’t want a system to be broken by North Korea, and then they don’t tell anybody else,” he adds.

Green, on the other hand, is worried that since cryptographic systems take years to update, discreet disclosures won’t be enough. “I definitely think that we don’t have any idea what’s coming,” he says.