Back when I did my BSc (Hons) in physics in 2003 we had a quantum physics/relativity lecture where we explored 4 different possibilities for beating the speed of light limit. On close inspection none of them was actually capable of beating the speed of light. If I recall correctly the third of the possibilities that we covered was quantum entanglement (we also looked at an interesting situation where the forefront of a propagating wavepacket can actually travel slightly faster than FTL, but the peak of the packet (which triggers the passing of information) was limited to the speed of light).
On the question of how Quantum entangled states work:
Firstly, we know from experimental physics that entangled states and FTL correlation do exist - "spooky action at a distance" to quote Einstein is in fact a real phenomenon. An experiment by a group in Geneva, Switzerland reported in Nature in 2008 determined that the "speed" of influence of quantum entanglement correlations has a minimum lower bound of 10,000 times the speed of light. (
http://www.nature.com/nature/journal/v4 ... 07121.html).
At the heart of entanglement is the fact that measuring an unknown quantum state will change that state in some way. The quantum states of two (or more) separate objects can become linked together in such a way that they must be described by a combined quantum superposition, not as individual objects. Once entangled the act of measuring the entangled quantity of one object will affect the other at FTL. This does not however mean that information can be passed instantaneously using that method.
The first problem with using entanglement to transfer information is that the entangled particles have to be separated - eg Alice gets one and Bob gets the other. Unfortunately the speed of light limit applies here - there is no way to move the two components of the entangled state apart from each other faster than the speed of light.

Also once you measure the state, the entanglement is over permanently so you can't simply reuse the same particles over & over again.
The next problem is that you can't simply encode data directly into the entangled states. For example, lets look at a typical protocol using entangled states.
The Ekert (E91) scheme uses entangled pairs of photons. The photons are distributed so that Alice and Bob each end up with one photon from each pair (but remember, they can't be distributed faster than the speed of light). The entangled states are perfectly correlated in the sense that if Alice and Bob both measure the polarization state of their particles, they will always get the same answer with 100% probability. Thus if Alice measures with a vertical polarizer and gets a photon out, Bob will too (a vertically polarized photon). If Alice measures with a vertical polarizer and gets no photon, Bob will too (a horizontally polarized photon). The same is true if they both measure any other pair of complementary (orthogonal) polarizations.
You might think that this means that once the system was set up (at speed of light transmission rates) Alice and Bob could later on start sending information at FTL - but you would be wrong. The problem is that the particular results are completely random - it is impossible for Alice to predict if she (and thus Bob) will get vertically polarized or horizontally polarized photons for any given entangled pair. Whenever you measure a quantum state it will randomly 'collapse' into one of the possible eigen states, but it is impossible to know which one it will be in advance. Measuring one member of the pair tells you what the state of the other particle would be if you measured it, but the results themselves are random. Bob and Alice can measure particles until the cows come home but they cannot pass information in this way.
What about Alice telling Bob before he leaves that they will use a code where the very fact that his particle has become polarised is the signal to do something (eg to buy 1000 shares in Vulpine Industries)? This still won't work because there is no way for Bob to know when to measure his particle. He cannot know if the photon is polarised UNLESS he measures it, and if he measures before Alice does then HE will be the one responsible for collapsing the entangled state, not Alice.
Even though you can't beat the speed of light this way, you can still make use of the correlation effect of entanglement for other purposes. For example, Alice can start measuring randomly using both a vertical polarizer and a circular polarizer, and Bob can do the same. They can then send each other a list of which polarizer they measured with at which time (over a speed of light link), and based on the results they can select out a random list of photons for which they both did the same measurement. This random string of 1's and 0's can then be used as a key for information encryption - and the benefit is that by looking at the list they can tell whether their key was intercepted by an eavesdropper! (see Quantum Encryption on wikipaedia for further info on how this works). Other applications of entanglement are superdense coding and quantum state teleportation.
