I'm all for regulating AI, but banning anyone from making superintelligence is a bit odd to me. It's hard enough to define superintelligence, so making a ban around it seems fraught. One could argue that current models are already superintelligent when it comes to coding. Do they have to be better than every single human to be superintelligent, or just your average person in that field?

Not to mention that China won't do this, so they would go on and develop it on their own.