The crowd control and crime prevention issue has been portrayed as very complex due to the urbanization involved, particularly complicated and high-density settings. Relying on this fact, this paper explores and presents a novel AI-based surveillance system utilizing YOLOv8 that identifies objects or anomalies in real-time as it addresses the presented challenges. This reduces human dependency to about 60% to 70% through automation of important functions such as weapon detection and crowd analytics, which is amazingly accurate at 96.3%. The technology will further enhance predictive analytics and enable proactive safety measures by using edge computing for low-latency processing that ensures quick and effective reactions to potential threats. The system further works well in different settings; high-density and low light are included, all because of scalability and adaptability. This innovation advances public safety management by changing surveillance from being reactive to being proactive, which gives a powerful answer to urban problems in modern times.