During major natural disasters like extreme weather events and wildfires, it is very important to notify citizens when the danger level exceeds certain thresholds. However, defining these thresholds in an entirely discrete manner is challenging, despite their major importance. Conversely, triggering an alarm without sufficient justification can be costly for the state and unnecessarily disrupt the lives of citizens. In this regard, employing a more flexible approach, such as fuzzy logic, in the context of artificial intelligence empowerment of the whole process, can provide rules to mitigate the discreteness of alarm thresholds. Failing to raise an alarm, when necessary, can result in significant human and property losses. In this article, we present a new approach that utilises fuzzy logic to classify danger thresholds in a smoother and more effective manner for the civil protection. The paper offers valuable insights into the practical implementation of Fuzzy Logic and the entire life cycle of the proposed Fuzzy logic-based approach. This approach is designed to be iterative, involving key stakeholders to ensure the validity and reliability of the results.