I would say that there is generally a notion that a machine should be designed in a way that minimizes harm even when it is misused, especially if it is intende

I would say that there is generally a notion that a machine should be designed in a way that minimizes harm even when it is misused, especially if it is intended to be used by the general public. I don't know if you agree with that, but I would say OpenAI seems to, at least superficially.