Differential Technological Development

An AI safety framework concept proposed by Nick Bostrom, which advocates prioritizing the advancement of technologies that reduce overall risk—such as alignment research—while slowing the development of technologies that amplify risk before adequate safety mechanisms are in place.