Recent developments in the field of artificial intelligence have provided a wide range of novel tools and capabilities to strengthen the crime prevention framework. Among the most significant of these tools are AI-based video surveillance systems, which, through the integration of closed-circuit television (CCTV) cameras and advanced image-analysis algorithms, play a prominent role in the management of public spaces by means of situational crime prevention. By employing technologies such as facial recognition, behavioral and emotion analysis, person and object tracking, and the detection of collective anomalies, these systems possess substantial preventive potential with regard to various types of crime, including theft, vandalism, terrorist offenses, child abduction, violent crimes, and others. This article, adopting a combined technical and legal-analytical approach, examines the preventive capacities of AI-based video surveillance systems. After identifying and analyzing the associated legal challenges, it assesses each of these issues in light of the risk-based approach, as embodied in the first binding regulatory framework in this domain, namely the European Union Artificial Intelligence Act of 2024. The findings of this study indicate that the deployment of such technologies, while reducing opportunities for criminal conduct, may constitute an important instrument of crime prevention, provided that legal, structural, and operational requirements aimed at safeguarding human dignity as well as individual and collective rights and freedoms are duly observed by the competent authorities.