Ideally, sufficiently powerful AI would not be created unless the necessary safety mechanisms are established.

But also, that’s a different kind of asymmetry?