Predictive Policing is a computerized algorithmic processing system that uses a database and statistical analysis to predict a future criminal event. System Little used in our country, but which is already in an advanced stage of operation in developed countries such as the United States of America and China. The system, which works based on criminal analysis, manages to foresee and prevent crimes, avoiding offense to legal interests and the consequent reduction of violence. This technology that has been implemented as a crime prevention strategy, used by the Brazilian police, needs fine and systematic adjustments. Even if in slow steps, actuarial criminal policies see this computer modeling system as an important contribution to crime prevention. Parallel to this technological advance, we have a growing legislative movement in the protective dogmatics of fundamental rights regarding privacy, intimacy and data protection. Thus, for a legal feasibility of this predictive policing system, its functioning must be analyzed in correlation with the constitutional rights and guarantees involved in the process. In this dissertation, we propose as a challenge, first, to collect reliable information about the technological programs of criminal prediction made available to public security and the technical-legal aspects involved. In a dogmatic-legal approach, we analyze the constitutional rights related to the operationalization of the system. Issues related to algorithmic discrimination and social inequalities are sensitive topics addressed in the study, as the predictive policing system, when poorly structured, can exacerbate social problems and be inefficient in fighting crime. Its source of data, when poorly researched, can bring about social discrimination and direct a policing that is born biased from the beginning and applied with partiality. The very functioning of the system reveals problems related to the management of personal data, intimacy and privacy, when they go beyond the limits of reasonableness. Another aspect is related to anticipated culpability, as the system classifies suspects and dispatches police officers to approach them even before any criminal activity occurs, only guided by a technological device of artificial intelligence, based on statistical analysis. The theories involved behind predictive policing software can still bring ideologies according to the line adopted by the system developer, with arms, for example, in zero tolerance policies and capitalist models of managing systems. In an even more hidden face, we can observe with a keen eye, aided by the work Discipline and punish: birth of the prison by Michel Foucault, that the State sometimes uses the discourse of combating crime and inspection to carry out social control and dominance of the masses. As the state is not obliged to be isolated from the social system, without taking advantage of technological advances, predictive policing can be an efficient tool in reducing crime and violence as long as its operation is compatible with the legal system and does not work to feed discrimination society, deepen inequalities and distribute injustices.