If machines have the ability to make decisions and take actions autonomously, they can therefore make morally unacceptable choices: this fact raises issues never faced before. There are two main approaches to the development of an ethics of intelligent systems: Ethics by design, in which norms are inserted so that the system behaves in an appropriate way under the ethical profile, and Ethics by learning, where intelligent systems learn, using an enormous amount of data on human behavior, how to become a moral agent. Whatever solution is pursued, the development of an ethics of AI is a complex and urgent process, that requires a collective commitment from researchers, developers, politicians and citizens.