"The practical applications of quantum encryption may be getting closer. A paper published in Physical Review Letters by Vedran Dunjko, Petros Wallden, and Erika Andersson presents a way to use Quantum Digital Signatures without requiring long term quantum memory.
provides a summary:
Quantum digital signatures (QDSs) allow the sending of messages from one sender to multiple recipients, with the guarantee that messages cannot be forged or tampered with. Additionally, messages cannot be repudiated; if one recipient accepts a message, she is guaranteed that others will accept the same message as well. While messaging with these types of security guarantees are routinely performed in the modern digital world, current technologies only offer security under computational assumptions. QDSs, on the other hand, offer security guaranteed by quantum mechanics. All of the variants of QDSs proposed thus far require long-term, high quality quantum memory, making them unfeasible in the foreseeable future. Here, we present a QDS scheme where no quantum memory is required, which also needs just linear optics. This makes QDSs feasible with current technology."
[Ed. Note] The Physical Review Letters link has all the fun details, but Phys.org provides a more understandable article for the layperson.
The quantum guarantees are based on the fact that you cannot measure an unknown quantum state without disturbing it. That is, if somebody would try to eavesdrop your quantum messages, you'd notice it. Given that the quantum communication is always just used to generate one-time pads, being able to detect eavesdropping means that you can avoid using OTPs which are known to the attacker.
Having said that, the security of course also depends on the attacker not being able to learn about your code any other way nor getting control of the output of your device (for example, there have been attacks on quantum encryption devices where an extra laser was used to find out the orientation of polarizators, or to blind the detectors with strong lasers and make them output whatever the attacker wanted.
As always, any scheme is only as secure as the weakest link. And that weakest link may well be the hardware which runs the quantum protocol.