Why should it mean that? - Can you give a more precise version of your question?
In case you refer to entropy as a indicator for "untidiness" in a system, the anser is: no, they don't brake it.
You cannot apply the classical term of entropy to macroscopic systems like balls, it only holds for a very large number of microscopic particles. The reason for this is that entropy is connected with the number of microscopic states which can be used for creating your macroscopic state. Since you are already dealing with a macroscopic state, you are completely ignoring (in a physical sense) your microscopic state spaces. Therefore you have already excluded the entropy from your macroscopic system.
A system does not violate the Laws of Thermodynamics by reaching its state of minimum energy.
A bunch of balls in a bowl. Mix up and shake for a while. Then stop stirring, and immediately all of them get together nicely grouped at the bottom of the bowl... State of minimum energy.
First of all: The 2nd Law (dS >= 0) for allowed processes) holds for the (internal) entropy change of an isolated system.
In the case of the magnetic balls, the experiment should be carried out as follows:
1. You put the magnets with random fixed orientations into a thermally isolated container.
2. You wait until the temperature reaches its equilibrium value.
3. You remove the fixation of the orientations, so that the magnets can align as they wish and read the temperature.
If the magnets are close enough to each other, they will probably align spontaneously (higher order, less entropy). But at the same time, the temperature will increase (higher entropy). If you calculate the net effect for steel balls, you will find that the overall entropy increases.
The system of magnetic balls consists not of those balls only, there also is an inseparable magnetic field generated by them and extending in space far away from balls alone. There would be no such observed effect if the field is not there, right?