... and that ought to be very clarifying, b/c that's simply how the world works. But it feels like a lot of x-risk philosophy and research is framed around the hope of providing guarantees in the model weights, rather than appreciating that safety is a systems practice—just as it always has been.
