Sort:  

A concise way to frame yesterday's "stop AI" letter:
(a) Major countries would need to agree and then uphold that agreement

(b) Enforcement would likely require a global authority to verify compliance ("let me check your model and datacenters to make sure you're living up to your word..")

Following this line of thought leads to a global body with near-unchecked power that proponents would try to staff or other nations could influence, potentially encroaching on US sovereignty