The difference between cryptographic algorithms and quantum key distribution (QKD) is that the algorithms operate purely in the mathematical realm of information theory. QKD is a physical process. Thus, it can do things that algorithms cannot do.
Let's say Alice wants to send a message to Bob without Eve getting the information. If Alice uses a cryptographic algorithm to encrypt the message to Bob, and Eve intercepts the message, she may be able to undo the algorithm. However, if Alice has a way to send a message to Bob such that Eve cannot get ahold of the information that is transmitted, then Alice could be sent the message in plaintext and be perfectly comfortable that Eve can't get at the information.
QKD relies on the laws of quantum mechanics to create such a channel. In theory, it is perfect. So long as the laws of quantum mechanics hold true, QKD is safe. However, we have to remember that it is a physical protocol, not a mathematical one. The ability to implement it is completely dependent on the hardware available. I read a few years back that one of the QKD systems could be fooled by shining a flashlight down the fiber, which fooled the hardware implementation of the receiver into exposing the secret information in the next step. It relied on the physical limitations of the single-photon detectors used in the devices. ( L. Lydersen, C. Wiechers, C. Wittmann, D. Elser, J. Skaar, and V. Makarov, Nat. Photonics 4, 686)
These physical attacks should be thought of as akin to other physical attacks on algorithms. AES, for instance, is not considered broken. However, there are some very powerful side-channel attacks that can attack the hardware implementing the AES algorithm.