First, thanks for the link. It looks an interesting subject and in retrospect it's obvious that an atom in a field gradient will seek to minimise it's energy by moving to where the field is strongest, though I must admit it had never occurred to me.
The phenomenon is simply due to electromagnetism. Atomic energy levels will split in an electric field due to the Stark effect, so some levels will fall in energy while others will rise. If all the levels are occupied there is no net change, but if only some of the levels are occupied there will be a net decrease in energy. Since light is just an oscillating electric field, it causes a (oscillating) Stark splitting so it can lower the total energy of an atom. If the light intensity is completely even the atom won't move because the lowering of energy is the same everywhere. However if the light intensity is uneven (e.g. the usual 1/$r^2$ dependance on distance from the light source) then the energy of the atom is increasingly reduced as it moves towards the light source. Differentiate this to get the force, and you get the $1/r^3$ dependance mentioned in the article (for a first order Stark effect).
So it's not a new force, just a manifestation of electromagnetism, but it is an example of how physics can surprise me even after 52 years :-)