First off, consider an alternate solvent.  Paraffin waxes contain some degree of unsaturation which could result in a site for oxidation to occur.  Once this starts it could potentially be the reason for your runaway.  Something completely saturated like any of the straight chain alkanes might be a better albeit more expensive and more flammable option.  But that would give you the option of running the milling at a lower temperature.

Second, look at the surface area.  Have you attained your desired particle size without incident before or has this happened every time?  Part of the takeoff in reaction could be attributed to you actually attaining that magical particle size where the surface area is such that it acts catalytically to oxidize your wax.  Surface area most certainly has the ability to catalyze a reaction, for example hydrazine can explode in contact with ground glass because of the surface porosity.  Permanganate is a strong oxidizing agent after all, especially in the solid form.  Ironically enough solid - solid reactions are now being facilitated by ball milling on the industrial scale.  Beating an oxidizeable liquid against it no matter how inert it may seem could just lead to such a reaction.  There are several binary explosive mixture that can be created from an oxidizer and fuel so long as they are intimately mixed.  I mean, usually I only see reducing agents (Lithium shot, aluminum powder, lithium hydride, etc) milled in oils or waxes.

But enough with the nay-saying.  First off, in the above situations, even with the magical particle size being achieved, it only serves to lower the activation energy.  That being the case cooling the vessel further than you already do during the milling could potentially circumvent this problem altogether.  Now, your paraffin might freeze during this operation but so long as the cooling is done slowly enough to keep the chunks free flowing they will still continue to roll against one another in true ball mill fashion and continue to break down the particle size.  I don't know if you need to go as far as super chilling it with liquid nitrogen, from my other communications with you, you're running at a high enough temperature where there is plenty of room to go down in temperature, additionally your ball milling medium could become embrittled at those temperatures.

A second consideration would be to measure the pH of of your starting materials.  There is the potential that your paraffin could have some trace acid or base.  Permanganate is able to oxidize under both acidic and basic conditions so going for neutral would be best.  Problem is permanganate itself is the salt of a weak acid so it's basic (couldn't find any information on the pH of a standard solution of potassium permanganate so this is just chemistry logic at work, if you try to measure the pH of such a solution, don't use a acid base indicator as it will probably just oxidize, use a probe).  If an aqueous solution of your permanganate though was acidic then some base should probably be added during the milling to neutralize this acid.  Anyway, its a stronger oxidiant under acidic conditions so if your paraffin is contaminated it should first be treated with a weak base to bring it to neutral, don't use ammonia as this could potentially cause an explosion.  (To do a quick measurement to determine if there is any contamination of your paraffin, try an extraction with hot water and titrate)  I bring this up because of analogous issues in pyrotechnic mixtures containing sulfur, the elemental sulfur is often contaminated with acid due to reaction with atmospheric moisture and oxygen which in the presence of some oxidizing agents, notably chlorate, make the shelf life of these mixtures unpredictable.

One experiment you could do to troubleshoot the process would be to pull samples at various points in your process and hold them at the processing temperature.  These samples containing different particle sizes of your permanganate could be your control samples.  If after holding for an extended period of time none of them exotherm, especially any taken just before the mixture in the ball mill takes off, you will have a good indicator that the it is less a case of any acid/base reaction at work.  Instead it would reduce the variables to a magic particle size (which would loose some credibility) to the actual smashing of the milling medium against one another giving the activation energy to the reaction.

In summary, try running it cooler as you suggested.  Investigate the possibility of a different solvent.  Also check the starting materials for acid/base content that could potentially lead to the reaction.  As with any real life case it could be any one of these, any combination of these, or something else all together.  Hopefully thought this will help to act as a guideline for other considerations in your processing.