idk i think theyre less idealistic and more monopolistic.

if they think their model is the best at making things safe, then their model should be helping train at any chance, to embed the same safety into new models