cross-posted from: https://lemmy.zip/post/24515831
The research team, led by Wang Chao from Shanghai University, found that D-Wave’s quantum computers can optimize problem-solving in a way that makes it possible to attack encryption methods such as RSA.
Paper: http://cjc.ict.ac.cn/online/onlinepaper/wc-202458160402.pdf
Follow up to https://lemmy.ca/post/30853830
Does this attack scale linearly with key size?
Using the D-Wave Advantage, we successfully factored a 22-bit RSA integer, demonstrating the potential for quantum machines to tackle cryptographic problems
That attack is a threat only if it scale better than existing attacks.
The headline should mention that they’re breaking 22-bit RSA, but then it would get a lot less clicks.
A different group of Chinese researchers set what I think is the current record when they factored a 48-bit number with a quantum computer two years ago: https://arxiv.org/abs/2212.12372
I guess the news here is that now they’ve reached 22 bits using the quantum annealing technique which works on D-Wave’s commercially-available quantum computers? That approach was previously able to factor an 18-bit number in 2018.
🥂 to the researchers, but 👎 to the clickbait headline writers. This is still nowhere near being a CRQC (cryptanalytically-relevant quantum computer).
Quantum computers were magical when they came out. The magic wore off and now they are… cracking tools?
So if you are genuinely worried about this, don’t.
First because, as numerous persons already clarified, researchers here are breaking deprecated cryptography.
It’s a bit like taking toothpicks and opening a lock while the locks used in your modern car is very different. Yes, it IS actually interesting but the same technique does not apply in practice, only in principle.
Second because IF in principle there IS a path to radically grow in power, there are already modern cryptography techniques which are resistant to scaling the power of quantum computers. Consequently it is NOT just about small the key is, but also HOW that key is made, what are the mathematical foundations on which a key is made, and can be broken.
Anyway for a few years now there has been research, roughly matching the interest in quantum computers, to what is called post-quantum encryption, or quantum resistant encryption. Basically the goal of the research is to find new ways to make keys that are very cheap to generate and verify, literally with something as cheap and non powerful as the chip in your credit card, BUT practically impossible to “crack” for a computer, even a quantum computer, even a powerful one. The result of that on-going research are schemes like Kyber, FALCON, SPHINCS+, etc which answer such requirements. Organizations like NIST in the US verify that the schemes are actually without flaws and the do recommendations.
So… all this to say that a powerful quantum computer is still not something that breaks encryption overall.
If you are worried TODAY, you can even “play” with implementations like https://github.com/open-quantum-safe/oqs-demos and setup a server, e.g Apache, and a client, e.g Chromium, so that they can communicate using such schemes.
Now practically speaking if you are not technically inclined or just want to bother, you can “just” use modern software, e.g Signal, which last year https://signal.org/blog/pqxdh/ announced that they are doing just that on your behalf.
You can finally expect all actors, e.g hosts like Lemmy, browsers like Firefox, that you use daily to access content to gradually both integrate post-quantum encryption but also gradually deprecate older, and thus risky, schemes. In fact if you try to connect today to old hardware via e.g ssh you might find yourself forced to accept older encryption. This very action is interesting because it does show that over the years encryption changes, old schemes get deprecated and replace.
TL;DR: cool, not worried though even with a properly powerful quantum computer because post-quantum encryption is being rolled out already.
So it factored a trivial (22 bit) RSA key.
1024-bit RSA has been deprecated for years. 2048-bit RSA is the recommended minimum.
Interesting. Not quite worrisome.