Because most of the so-called consequences of malevolent AI which aren't some variety of killdrone or workforce replacement aren't actually that bad, or at least are vastly preferable to the measures which would be necessary to prevent them.

The typical arguments are that AI art and deepfakes will destroy 'art' and the credibility of the news, with the only ways of avoiding this being to butcher privacy on the internet and pass extremely far-reaching copyright laws.

The reality is, giving everyone access to the equivalent of a Hollywood special effects studio and actors will create a fucking renaissance and there's not much AI could do to drive news credibility any lower than human reporters already did. ^("Iraq has weapons of mass destruction." "Anyone who loses their job because of the new trade deal we just made will be retrained and get a better one." "We're not spying on our own citizens." "We'll be welcomed as liberators." "But) ^(this) ^(group of insurgents are Moderate Freedom Fighters™, not bloodthirsty jihadist terrorists." "Jeffrey Epstein killed himself.")