I think the Darwinian logic of reality might make this hard. If society A and society B are both developing AI and one of them stops in order to protect humans, society B may continue to develop AI and then it might either outcompete society A economically to the point of reducing it to poverty (it is theoretically possible to take most of another society's market share in something by only slightly outcompeting it in price or quality), or it might even outright conquer society A.
A solution to the problem needs to address this issue somehow.