Historically, technologically advanced nations have worked together to develop some limited governance over sensitive technologies. It's an imperfect system, but tends to be better than a free-for-all.

Counterexample: Atom bombs, ICMBs, spy satellites.

You literally just named the actual examples.

MAD is the opposite of cooperation.

The cold war involved numerous bilateral agreements, accords, understandings, etc to prevent a nuclear war.

Or they are the first countries falling to the superintelligent AIs they themselves have developed... :o

... with the rest of the world following shortly.

I think this argument is frankly quite ignorant. The only adversarial nations that maybe have the capability to build these systems are Russia and China. Russia just got sanctioned to hell and the Chinese government is limiting their AI development more than we are. By far the most gung ho country about developing AI is the United States and its tech sector. Bringing up the CCP strawman is just an excuse for less government oversight over what the industry itself has been saying is a dangerous technology for months now.