Maybe technically, but it's rather unlikely you will get good results. If we take two potentials, and one has interactions for A-B and the other for CD (assuming a system with A+B+C+D) you will not have any forces at all between A-C and the other missing pairs. If, on the other hand, you have AB and AC, it will be confusing how to handle AA interactions which most likely are essential - from which potential should they be taken (likely those parameters are not identical)?
Feels like a good opportunity to try MTPs instead :-)
