habryka comments on New report: “Scheming AIs: Will AIs fake alignment during training in order to get power?”