How might we make
our AI services as
irresponsible
as possible?

I ran a reverse brainstorming workshop with the full AI squad. We started with the worst possible answers — then flipped them to help create our guiding principles.