The scope of analysis is confined to four safety domains—fire, crime, pedestrian, and flood safety—and methodologically covers 177 Acts and 107 administrative regulations using AI-based processing. A review of previous studies reveals that while most existing research focuses on single safety domains or governance reform, the present study is distinct in that it applies a hybrid ontology–LLM mechanism to diagnose normative contradictions that emerge when multiple safety regulations overlap within a single spatial environment.