We make a simple case for regulating advanced AI past a certain point. Part 1.