The rise in the global population, the effects of climate change, and the growing urbanization in regions prone to natural hazards are some of the factors contributing to the exponential increase in the economic and human losses due to natural hazards. In response to this global challenge, several international agendas such as the Sendai Framework for Disaster Risk Reduction, the United Nations 17 Sustainable Development Goals, and the Paris Climate Agreement specifically demand a better understanding of disaster risk and impose specific risk reduction targets for the upcoming decades. Yet understanding, quantifying, and monitoring risk remain a challenge, particularly in areas where traditional risk data are limited. In the last decade, scientists, risk modellers, and data analysts have explored cutting-edge technologies such as artificial intelligence and data sources with unprecedented detail and spatial resolution to improve the reliability, accuracy, and geographical coverage of disaster risk assessment studies. The huge rise in popularity of these methods also comes with potential risks. This special issue fills an urgent need to properly understand the potential for AI and machine learning in the field of disaster risk analysis, communicate potential pitfalls and limitations, and provide examples of best practice. This special issue presents some of the recent advances and applications of machine learning in natural hazards risk assessment, covering the following topics:

• exposure data collection and automatic building classification;
• risk mapping and damage assessment;
• development of hazard intensity and vulnerability models;
• monitoring population and urban growth;
• post-disaster damage and loss detection;
• modelling of socio-economic vulnerability;
• when to use AI or machine learning vs. when to use conventional disaster risk analysis models;
• data, models, uncertainty, and performance metrics;
• ethical considerations in the use of AI for disaster risk modelling.