In 2016 police forces on the Belgian coast began to apply predictive policing software to their operations, echoing a larger movement and investment within the Belgian government to build centralized cloud-based police data infrastructure (Algorithm Watch 2019). The chief commissioner claims that the project has been successful and correlates a 40% drop in crime to the start of the project (idem). There are hopes to further bolster the system by connecting it to Automatic Number Plate Recognition cameras (idem). Belgian federal and local police have called for the expansion of predictive policing, believing it to be a critical tool that not only saves time and money but prevents and stops crime. According to a spokesperson of the Federal Police, these tools and systems are currently being engineered for the national scale with data sets being collected and aggregated from police forces and third-party sources (idem).
Similar to other predictive policing software, the model relies on historical data sets to predict crime across geographic areas. Scholars and data ethics consultants question the legitimacy of these systems as they have been shown to exact disproportionate costs on marginalized communities and reify social and political inequalities (O’Neil 2018). Nonetheless, the companies that market predictive software prefer not to disclose the internal operations of their algorithms. American firm PredPol and Dutch company Sentient cloak their models in secrecy, refusing to divulge what criteria and inputs are considered.