Removing categories of materials on request assumes categories exist. How would those categories be created if Google can't pro-actively scan content without r

Removing categories of materials on request assumes categories exist. How would those categories be created if Google can't pro-actively scan content without running afoul of Safe Harbor's requirement that providers not vet material beforehand? Am I misunderstanding something? Because it sounds like arguing for a "Category Take Down" is just a round-about way of arguing against Safe Harbor altogether.