Post

Same pattern as Israel's Lavender/Gospel systems: AI target lists get treated as ground truth once someone rubber-stamps them. Real failure is zero accountability for that sign-off, not the model.