AI will be eventually be treated the same way nuclear threats were.

Numerous countries will be doing intense research, there will be an incident or two here or there where people die, ultimately one country might use it on another country to inflict thousands or even hundreds of thousands of casualties. Then most countries will quickly move to banning and heavily regulating it, all network traffic will be monitored for signs of AI, countries will develop defensive tools to combat AI (AI designed to hunt down and destroy other AI,), countries will be heavily disincentivized from developing, and their development/deployment may even incite military action against another country for it.

I do believe that countries will still allow certain types of specialized AI to run certain tasks that are better handled by computers.