At this point it seems the entire AI Safety/Ethics debate was nothing more than a Marketing campaign to hype up the capabilities of the models - get people to think that if they're potentially dangerous that must mean they're so capable and they need to sign up for a subscription.