It appears that the differences in timing and intensity of two upper-level troughs have direct influences on the position of the ridge over the western Atlantic Ocean. The first feature is a shortwave trough traversing western Canada.
The GFS depicts this feature to be more amplified and slightly faster than the ECMWF. If this were the case, the larger, broader piece of the trough to the south would intensify due to more intense cold advection from the north. The result would be a more significant force pushing against the unseasonably strong wall of a ridge over North America, shifting it eastward. With the ridge shifting eastward, the clockwise circulation can pick up Florence and direct it toward the north, resulting in a chance for the storm to remain just offshore or making landfall in northern North Carolina or Virginia.
These same upper-level features are weaker and slower on the ECMWF. Thus, the ridge remains centered further west, keeping Florence further south and west, resulting in a landfall somewhere in the Carolinas.