Comment by atleastoptimal

IMO the best path is global regulation on frontier AI development with high penalties for defection that, detectable and enforced by member states, independent agencies and broad public support.