4
$\begingroup$

Not sure if this is a stupid question, but can quantum chips be spied on?

I heard that quantum particles(?) can be "entangled".

So is it possible that a chip manufacturer (who wishes to spy) can pretty much entangle all the particles on the chip so that what ever processing goes on inside that chip is mirrored miles away by the spy manufacturer?

$\endgroup$
  • $\begingroup$ No more than the ISPs. Even the quantum teleportation protocol requires classical data. The chip would still need to be connected to the internet somehow. $\endgroup$ – Sanchayan Dutta Dec 11 '19 at 18:59
  • $\begingroup$ Ok my information might be bad then. I thought the particles could communicated without the internet. Is it possible the spy particles could be at the manufacturer, and their states will be mirrored there without the internet? Then their states can be converted to normal classical data.. ? If no, then thanks much for the response :) $\endgroup$ – swift nub Dec 11 '19 at 19:04
6
$\begingroup$

Quantum states are highly unstable and subject to noise, so with current levels of technology it is unlikely that this entanglement would last for any useful amount of time.

But let's assume that you have two entangled fault-tolerant qubits (one owned by a malicious manufacturer), and that they are in superposition (in the Z-basis): $$ \frac{1}{\sqrt{2}} (|01\rangle + |10\rangle) $$

Where the first qubit is owned by the manufacturer. These are perfectly anti-correlated, meaning that a measurement of 0 on the first qubit causes the second to be measured as 1.

This may seem promising if you are a nefarious manufacturer; however the moment the user mesures their qubit, they collapse the superposition and the two qubits are no longer entangled, you have two separated states $|0\rangle$ in the user's computer and $|1\rangle$ in the manufacturers. You cannot bring them back into superposition without some interaction between the two qubits again.

Quantum computations are repeated many times to get the expected value. With this scheme you get only one of the (probably millions of) repeats.

And as @Sanchayan Dutta mentions, the choice of basis is important (this is what we communicate classically in teleportation), if the user were to rotate their measurement basis randomly before performing the measurement, the manufacturer would be unaware of this. The manufacturer's qubit would be in an eigenstate of the basis, but if they then chose to measure using the Z basis, it would collapse into an eigenstate of that basis.

$\endgroup$
  • $\begingroup$ because this has 3 upvotes, i'll mark it as correct. But to a normal human like me, i have no idea what you just said. $\endgroup$ – swift nub Dec 11 '19 at 21:58
  • 1
    $\begingroup$ @swiftnub We're all normal humans here. If you want to learn the subject properly pick up a good textbook. The basics of quantum computing are pretty accessible to even motivated middle school kids. :) $\endgroup$ – Sanchayan Dutta Dec 11 '19 at 22:11
3
$\begingroup$

Yes, and no. (Helpful answer, I know)

If you rely on buying the entire package from a single supplier, then yes. This doesn’t need to have anything to do with quantum at all. A quantum computation starts and ends with normal classical bits on a normal classical computer. If that was supplied as part of the package, it can easily have some malware built in that transmits all those classical inputs and outputs, which is sufficient for the supplier to entirely replicate everything the user got.

However, if you’re willing to separate the components a little, you can get something far, far stronger. Let’s say that, as a user, I can supply my own classical computer that I am confident is free from tampering, and I can produce my own single-qubit quantum states. I can perform a protocol called “blind quantum computation” to interact with the quantum computer. Even if the quantum computer part is completely compromised and untrustworthy, they learn essentially nothing about the computation you’re running beyond some basic property of how large the computation is. This was originally designed so that you don’t even have to own the quantum computer yourself. You can leave it with the manufacturer, who can continue to be as naughty with it as they want, and they still learn nothing!

$\endgroup$
  • $\begingroup$ as a side note, i know it doesnt need quantum for spying. Its just that with quantum (if it could be done), it would be practically undetectable. I know that humans are extremely creative, so i just wanted to know if using all the creativity/clever tricks, if this could be done. Based on both your answers, the answer is no, so thanks for letting me know. I was curious :) $\endgroup$ – swift nub Dec 12 '19 at 12:48
3
$\begingroup$

So is it possible that a chip manufacturer (who wishes to spy) can pretty much entangle all the particles on the chip so that what ever processing goes on inside that chip is mirrored miles away by the spy manufacturer?

No, not really.

The big reason this can't be done is that quantum entanglement doesn't allow for communication. Suppose that there's a particle A in New York, and a particle B in London, and that the particles A and B are entangled. Then, if I manipulate particle A, this will have no observable effects whatsoever on particle B. The people in London will have no idea what I'm doing in New York.

This fact is called the no-communication theorem.

Of course, that's not to say that it's impossible for a chip manufacturer to spy on the users of the chip. For example, the manufacturer could put in a hidden radio transmitter. But they can't spy by means of entanglement.

$\endgroup$
0
$\begingroup$

Yes, you can in principle spy on a computation - albeit using more subtle/gentle methods than entangling everything -, but you can prevent it.

The setting is that of a client who whishes to perform a computation on a server. If nothing specific is done, the server will know the computation being performed (ie the algorithm) because you will send a sequence of commands to it. It might also spy on you data (easy if the information is classical, harder if it is quantum, but not impossible).

So basically, delegated quantum computation is not blind to a malicious server.

But you can prevent it using blind delegated quantum computing schemes. These schemes use measurement based calculus to prevent the server from learning anything from the algorithm being done and the data it ia applied to - except the length of the compuation and the size of data. One scheme for doing so is BFK09 https://arxiv.org/abs/0807.4154

You can improve on this result by adding verification. That is you prevent a malicious server to fiddle with the compuation as you verify it is doing the right job (in case he doesn't comply with your instructions, you'll catch him with high probability).

$\endgroup$
0
$\begingroup$

No,It is impossible to copy the quantum information according to No-cloning theorem enter link description here. If anyone try to copy or measure the states of qubits the previous information will be lost.So no one can spy. This is the backbone of the Quantum-cryptography

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.