The relationship between AI and power is a robust case of the general relationship between knowledge and domination. At present, countries and corporations are competing for the best solutions, and the weight of business—in an increasingly less regulated world—is progressively greater. Under the guise that the private do better and cheaper than the State, the business logic of profit invades critical areas such as Defence, Health, Justice, Education and Social Security. The transformation of each citizen into a user or customer with a defined potential is advancing rapidly. In this circumstance, it is urgent to regulate. While it is true that a certain level of competitiveness is healthy and beneficial for all, it is also clear that pure competitiveness leads to knowledge protection logics and to the introduction of solutions in the market without proper safety testing. In short, a general attitude of not caring about the means that serve the ends prevails. Here we argue that there should be regulatory institutions and control processes aiming at developing a benevolent AI, one at the service of society and of knowledge, which is fair for all.