In this article, I conduct an exploratory study based on first-hand information from developers and police agencies to outline the current implementation of predictive policing in Italy. While highlighting several actors’ lack of responsiveness and transparency, the study elucidates a predictive tool’s features and offers insights into the Italian context’s peculiarities. Key aspects include the arguments for secrecy advocated by police agencies, the plans of the Department of Public Security for upcoming attempts to relaunch predictive policing, and the state of the predictive policing market, which, at the moment, appears dominated by buyers’ scepticism and companies’ struggle to still build their business image. Additionally, field-based reconstruction provides an overview of the main political and legal issues, which I deliberately broach from a non-privacy-centred perspective. Based on this analysis, I challenge the secrecy arguments and critique the dominant regulatory approach that focuses on ‘profiling-like’ outputs as the ultimate test for the lawfulness of AI-led predictions in law enforcement.