There were reports of MHD simulations of test accretion discs in the "magnetically arrested" or "magnetically choked" state with efficencies up to $300\%$ (e.g. here). I.e., the outflow of energy turned out three times the matter inflow. In fact, there are sources for which the efficiency is estimated to be significantly more than $100 \%$ and energy extraction from black holes seems to be a good explanation. Blanford-Znajek, Penrose, and recently also Banados-Silk-West, all these analytically formulated processes suggest that it should be possible to extract energy at the cost of the black hole mass.
However, both the exact and numerical computations are always done in the test mode, the particle/fluid/raditation takes away the extra energy but does not affect the background space-time. From the very principle of the process, only non-test computations can show whether the energy extraction really happens or not. The rest is only conjectures. I am quite positive that no one has studied even the linear influence of the matter on the background of a Kerr BH in this context and would not expect numerical relativist to be on it (they are nowadays mostly concerned with gravitational waves).
But yes, especially the mentioned MHD simulations find energy extraction in plausible astrophysical scenarios and it would seem improbable that such a strong and unambiguous effect would be changed radically by non-test behaviour.