Brilliant (literally) hack

[unixronin]

Via Bruce Schneier:  A team of cryptologists at the University of Trondheim, Norway, have developed a successful attack that works against both currently-deployed existing quantum cryptographic systems, IDQ and MagiQ.

The attack, which allows an eavesdropper (traditionally "Eve") to completely recover all quantum-encrypted data sent over the link, completely invisibly to the recipient (traditionally "Bob"), is brilliantly simple:  Eve simply blinds Bob's quantum detector with a 1mW laser, preventing it from operating as a quantum device capable of detecting single-photon polarization, then intercepts all the entangled photons and reads them herself.  Eve then resends every 1 bit to Bob as a bright laser pulse, which Bob's detector, blinded for quantum events, responds to in classical mode and reads as a 1.  Bob's detector — and Bob — cannot tell the difference.

"We have exploited a purely technological loophole that turns a quantum cryptographic system into a classical system, without anyone noticing," says Makarov.

Quantum encryption has been widely considered, and widely touted as, unbreakable because the laws of physics guarantee that you cannot measure any property of a quantum system without detectibly disrupting the system.  Thus it has been taken as gospel that you cannot eavesdrop on a quantum communication channel without leaving clear evidence that you have done so.  This inspired piece of lateral thinking is a Kobayashi Maru strategy — faced with an unwinnable game, Makarov's team have simply changed the rules of the game to one that they can win.

Comments

[merlinpendragon]

I'm not sure which gets my attention more, the fact that they did this or the fact that in the explanation of what was done, they invoked a Star Fleet leadership aptitude test to identify a "no-win" scenario! COOL!

[unixronin]

Well, actually, that was my analogy. :)

[vatine]

Oh! I think I see, it's just a temporary overwhelming rather than knocking the sensor (partially) out forever!

Of course, any attempt to detect the photons elsewhere (to see the constant rate) might well collapse the effects used for the quantum encryption, so I doubt it's trivial to detect this.

[unixronin]

[...] so I doubt it's trivial to detect this.

Well, no, actually the main reason this attack is not detectible at present is because no-one had thought of looking for it, since it was a previously unknown technique. Now that the technique is known, it ought to be possible to modify the systems to detect for such an attack by looking for the 'blinding' signal.

[unixronin]

quantum detector with a 1mW laser