The Day a Quantum Computer Learns to Pick the Internet's Locks | STEMBridge Learning

The Lock, the Algorithm, and the AI Quietly Picking It

Quantum Computing’s Collision Course with Your Encryption

Imagine checking your bank account one morning and finding that the encryption protecting it has effectively stopped doing its job.

Your messages could be exposed. Digital signatures that once proved a document was authentic could become questionable. Government records and other systems that rely on public-key cryptography could suddenly have a much bigger problem than a software bug.

Nothing about the websites would necessarily look different. There might be no red warning screen. The lock icon could still sit quietly in your browser.

The difference would be that someone had learned how to pick the lock.

That sounds dramatic, but the lock in question is built from mathematics, and mathematicians have known for decades what kind of key might eventually be able to break it.

The strange machine behind the problem

To understand why quantum computers matter, you don't need a physics degree. You need to understand one small difference between a normal computer and a quantum one.

A classical computer uses bits. A bit is either 0 or 1. Quantum computers use qubits, which can exist in a combination of states through a property called superposition. Qubits can also become entangled, meaning their quantum states can be linked in ways that have no straightforward classical equivalent.

The analogy I find most useful is a coin.

A normal bit is like a coin sitting on a table: heads or tails. A qubit is more like a coin spinning in the air. Before it lands, you can't simply point at it and say, "That's definitely heads." Its state has to be described using probabilities.

That spinning coin isn't a magical shortcut for every computing problem. That's the part that often gets lost in the hype. Quantum computers are useful because algorithms can manipulate these quantum states in carefully designed ways, allowing certain mathematical problems to be approached very differently from how they are handled on classical machines.

And one particular problem is where things get uncomfortable for modern encryption.

Why RSA has survived for decades

RSA, named after Ron Rivest, Adi Shamir, and Leonard Adleman, was introduced in 1977. Its security rests on a deceptively simple mathematical problem: factoring large numbers made by multiplying two large primes together.

The system uses two mathematically connected keys. One is a public key, which can be shared. The other is a private key, which is kept secret.

Suppose two large prime numbers are multiplied together. Multiplication is straightforward. Going in the opposite direction and figuring out which two large primes produced the result is a radically harder task for a classical computer when the numbers become sufficiently large.

That asymmetry is the whole point.

RSA doesn't need to hide the public key. It can be sitting in plain sight. The security comes from the fact that an attacker who knows the public information should still face an impractical mathematical problem when trying to recover the secret information.

That's why RSA has become one of the foundational tools of public-key cryptography. It has been used for secure data transmission, digital signatures, TLS/SSL certificates, and parts of banking and other systems that need to establish trust over networks.

For roughly half a century, the basic assumption has held: classical computers are not good at efficiently factoring the enormous integers used by RSA.

Then came Peter Shor.

Shor's algorithm is where the lock starts to crack

In 1994, mathematician Peter Shor developed a quantum algorithm for finding the prime factors of an integer.

The clever part is that Shor's algorithm doesn't simply tell a quantum computer to "try lots of factors." Instead, it transforms the factoring problem into another problem that a quantum computer is particularly well suited to attacking: finding a mathematical period, or order.

The algorithm has a classical part and a quantum part.

First, the factoring problem is reduced to an order-finding problem. If the relevant conditions work out, finding that order gives the information needed to recover factors of the original number.

The difficult part is finding the order efficiently. That's where the quantum computer enters.

Shor's algorithm uses quantum order-finding together with the Quantum Fourier Transform, a mathematical operation that helps extract the periodic structure hidden in the quantum computation. Once that information is obtained, classical mathematics finishes the job.

The result is striking. Shor's algorithm can factor integers in polynomial time on a sufficiently capable quantum computer, compared with the much harder factoring problem faced by the best-known classical approaches.

And that is precisely why RSA is vulnerable to a sufficiently powerful quantum computer.

Not because quantum computers are simply "faster computers." They aren't. The threat exists because a quantum algorithm changes the computational difficulty of the specific mathematical problem RSA depends on.

There is even a simple Qiskit demonstration of the idea: give a quantum simulator a small number such as 15, run a version of Shor's algorithm, and obtain its factors. That's obviously nowhere near cracking real-world RSA, but it demonstrates the basic mechanism.

The gap between factoring 15 and factoring the enormous integers used in real cryptography is, however, enormous.

Where AI actually fits

This is where I'm going to separate what the sources establish from my own reading of where the technology is heading.

AI is not secretly sitting in a server somewhere cracking RSA today. That's not the interesting part.

The more interesting possibility is that AI and quantum computing could help each other develop.

Quantum computers are notoriously difficult machines to build and control. Their qubits are fragile, errors matter, and useful algorithms require careful circuit design. Machine-learning techniques could potentially help researchers search for better quantum circuits, tune systems, and improve approaches to error correction.

AI can also be useful for simulation. Classical computers struggle to model increasingly complicated quantum systems directly, so machine-learning methods could become another way of approximating or studying aspects of those systems. The goal isn't to make a classical computer secretly become quantum; it's to give researchers better tools for understanding and designing quantum hardware.

Then there is quantum machine learning, the emerging idea of using quantum computers as part of machine-learning systems. Whether that eventually produces major practical advantages remains an open question. It is a field of research, not a finished product waiting behind the curtain.

But this combination creates an interesting strategic problem.

Imagine that progress in AI helps researchers design better quantum hardware, while better quantum hardware eventually makes certain computations dramatically easier. The two technologies don't have to directly "merge" for their progress to interact. Improvements in one could accelerate work on the other.

That is my reason for thinking post-quantum cryptography deserves attention now rather than after the first RSA-breaking machine appears.

Cryptography has a long replacement cycle. Systems are built into browsers, servers, financial infrastructure, government networks, devices, certificates and software libraries. Replacing a cryptographic foundation isn't like changing your phone wallpaper.

If AI-assisted research helps compress the path toward useful quantum machines, the sensible response isn't panic. It's preparation.

No, your passwords aren't getting cracked tomorrow

There is a huge reality check here.

A quantum computer capable of running Shor's algorithm against real-world RSA at meaningful cryptographic sizes does not currently exist. The algorithm may have an extraordinary theoretical advantage, but practical quantum computing still faces the problem of building machines with enough useful qubits and sufficiently low error rates.

That's not a minor engineering detail. It's the entire mountain.

A demonstration on a small number is fundamentally different from attacking the enormous numbers used by real cryptographic systems. Quantum computers today remain limited and error-prone compared with what would be required for this particular threat.

So if you're reading this on your laptop, you don't need to change your passwords because a quantum computer is about to break into your bank account tonight.

The interesting question is not whether RSA is doomed tomorrow.

It's whether we should wait until the day it is.

The lock was never permanent

What fascinates me about this problem is that RSA isn't failing because someone found a silly mistake in the original idea.

It worked.

It has worked because the mathematical problem behind it has been brutally difficult for classical computers. Shor's algorithm doesn't make that history irrelevant; it shows that security assumptions can depend on the kind of machine doing the computation.

That's a bigger lesson than quantum computing itself.

Every generation builds technology around the machines it has. Then someone invents a different kind of machine and asks whether the old assumptions still hold.

For people my age, that shift is happening while we're watching.

The passwords, messages and transactions around us are protected by mathematics designed for a world of classical computers. Somewhere else, researchers are trying to build machines that calculate according to a different set of rules.

Maybe the quantum computer capable of breaking RSA is decades away. Maybe progress surprises us.

Either way, I'd rather be living through the period when we're replacing the locks than the morning when we discover someone has already picked them.

What do you think: should post-quantum cryptography already be treated as a priority, or is the current quantum threat still too theoretical to worry about?