This paper provides an introduction to the emerging research track of fairness in automatic classifiers operating on, typically, personal data. A fair classifier is characterized by equality of opportunity for the in- dividuals, i.e. the classifier is unaffected by individual characteristics it should ethically and/or legally be agnos- tic to. Such situations are potentially ubiquitous in the increasingly automatd digital society, e.g. in the context of automatic processing of load or job applications, or on on-line personalized information tools. The main goal of this research track are the translation of criteria from public policies into statistical criteria and their introduction into computerized regulation techniques. Besides detecting discrimination in a given decision pro- cess, recent papers propose interesting first approaches for enforcing fairness, through data processing techniques. Finally, we describne how, beyond the general goal of protecting individuals, data processing for fairness and for anonymity share several scientific problems and solutions.