That's a great question. In my analyses I found that to mitigate infections in close encounters, it is far more important to have a strong airflow stream that is directed into the interactions between people than a HEPA filter.
A HEPA filter is made of high density fibers and would produce too much air resistance to be of any use when we are trying to project an air stream up to a distance of 3 meters with a low-cost low-power household box fan. Otherwise, the solution would need an expensive powerful fan motor that would also cause too much noise to achieve that kind of high airflow rate at a long distance. This solution has to be affordable in poor countries and also has to be energy efficient.
Instead, the crosswind unit uses a combination of three medium-MERV filters that together have an effective efficiency of approximately MERV-12.5 [view here in Part 3]. So the tradeoff is that it takes longer to scrub a room to HEPA quality in exchange for a strong reasonably clean air stream that instantly deflects the infectious particles [view the difference between here and here in Part 3]. Therefore a medium-MERV (see Question Q15) filter set is sufficient to achieve the desired objectives. See also Question Q16.