Suppose we couple $N$ colors of fermions to an $SU(N)$ gauge field $A$, but instead of a Yang-Mills action, there is a BF theory that restricts the gauge field to be flat $dA+A\wedge A\equiv F=0$ (by using a Lagrange multiplier field called $B$, hence the name). In some sense the BF theory is like an 'asymptotically free' limit of Yang-Mills where the gauge coupling is taken to zero.
If we are considering the fermions to live in a topologically trivial space like infinite volume Minkowski space, what is the difference of this BF coupled theory from a theory of $N$ free fermions?
It seems there must be some difference in the sense that if we have a finite temperature we can have non-trivial flat gauge fields twisting around the thermal circle, and this should lead to different thermal correlation functions. Is coupling to the BF theory just equivalent to just filtering out the non-gauge invariant states from the spectrum of the free theory?