A series of warnings over the risks of artificial intelligence (AI) has created an urgent imperative for the AI race: it is not enough simply to build more powerful models; efforts must also focus on controlling the pace and potential dangers, ensuring that humans remain in command of the technology’s development trajectory.