In this article, a preliminary assessment of fire hazards in public buildings is given.and the effectiveness of artificial intelligence (hereinafter - AI) models in the forecasting process. According to the results, the artificial neural network model increased accuracy by 22-25% compared to classical statistical regression models, and the speed of early risk detection by 30-35%. It has also been scientifically determined that the possibility of real-time analysis can reduce the impact of the human factor and accelerate the evacuation process.