Artificial Intelligence in Construction Safety: A Decade of Progress, Challenges, and Future Implications

The construction industry is regarded as one of the riskiest industries in the world, due to the high accident rates and severe socio-economic implications. Technology has progressed incredibly in the last decade, with Artificial Intelligence (AI) based methods that offer the potential to advance safety within the construction sector through advanced hazard detection, prediction, and prevention. While there has been an undeniable surge in academic literature investigating AI in construction safety over the last decade, there remains a gap between the research level of AI tools and their deployment to support practical safety implementation, and no surface synthesis of common themes, emerging trends, methodologies, and contexts. This scoping review reports on 42 peer-reviewed empirical studies investigating AI in construction safety published from 2013 to 2025, following the PRISMA protocol. Overall findings exhibit deep learning as the primary method, and vision-based systems as the most prominent data type. Research largely remains in the domain of personal protective equipment (PPE) detection and compliance monitoring, including near-miss prediction, real-time deployment, and multi-sensor fusion. Most studies are conducted in Asia, with less emphasis on the presence of studies in developing countries. This review identifies technology, thematic, and contextual patterns and provides an integrated reference point for researchers and practitioners. It is noticed that an AI safety framework needs to be contextual, regionally adaptable, and codeveloped. Future directions should favour AI systems that are real-time, multi-modal, and scalable, including conducting more relevant lab research that can be used practically in the field. A multi-stakeholder, interdisciplinary effort will enable full utilization of the potential of AI for safety outcomes in different types of construction work.

Keywords: Artificial Intelligence, Construction Industry, Safety and Risk Management, Predictive Analytics.