Well, depending on what lab/industrial molecular sieves one is referring to, they might actually be zeolites. All zeolites are molecular sieves, though not all molecular sieves are zeolites.
Despite what the company claims, I strongly suspect that the mode of action for the K-Z zeolite is largely physical sequestration of PO4 in the molecular sieve mode of action instead of strict removal of PO4 by bacterial growth/skimming.
There are several reasons that lead me to this conclusion - if the mode of action was just bacterial growth/removal by skimming, it wouldn't be necessary to have a zeo reactor nor any zeolite in the system at all because any reef tank with a substantial quantity of LR has an enormous surface area for the growth of biofilms/bacteria that far exceed the surface area in a few hundred mls of zeolite. Moreover, mature biofilms on surfaces constantly shed into the water, so stirring a zeolite in a reactor isn't going to substantially change the amount of bacteria in the water available for skimming out.
Finally, if the primary mode of action of the zeolite was simply growth of bacteria, then it would not be necessary to replace the media.
However, if the zeolite is a correctly sized molecular sieve, than it will selectively absorb phosphate from the water, even in a seawater system with abundant competing ions. Under such circumstances, having to stir the zeolite on a frequent basis and having to replace it at some regular interval makes perfect sense. One would have to break up the bacterial biofilm on the surface of the zeolite on a frequent basis to ensure continued diffusion of the phosphate into the zeolite, and eventually the zeolite medium would become exhausted - just as it does in industrial processes that use molecular sieves as purification.