Are you suggesting OpenAI be forced to make everything public before they've figured out the safety systems, because you're afraid of the hypothetical unfolding if they don't? That sounds like a fear-driven response to me, and no way to propose good policy. Do correct me if I'm mischaracterizing.