The Italian city of Trento has launched an ‘eSecurity’ initiative with the support of various backers and partners, including the European Commission, the Faculty of Law at the University of Trento, ICT Centre of Fondazione Bruno Kessler and the Trento Police Department (AlgorithmWatch, 2019). The project describes itself as “the first experimental laboratory of predictive urban security”, and subscribes to the criminological philosophy of ‘hot spots’ where “in any urban environment, crime and deviance concentrate in some areas (streets, squares, etc.) and that past victimization predicts future victimization” (ibid). The system is largely modeled after predictive policing platforms in the US and UK, and employs algorithms that learn to identify these criminal ‘hot spots’ from analyzing historical data sets.
While Trento’s eSecurity initiative appears to be an objective method to predict and tackle crime, in turn, allocating police resources efficiently and promoting public safety, it actually falls into the same trap as the predictive policing systems that have come before it (O’Neil, 2018). Due to the disproportionate amount of data collected over petty crime (which is generally carried out in working-class/minority neighborhoods), such models tend to direct police surveillance to low-income and already over-policed communities. This instates a feedback loop that perpetuates existing social inequities (ibid).