The field of construction safety and data-driven solutions is moving towards increased adoption of artificial intelligence and machine learning techniques. Researchers are exploring the use of Vision Language Models (VLMs) for tasks such as detecting safety rule violations from on-site images, and developing open datasets to comprehensively evaluate and fine-tune VLMs. The creation of systematic catalogs of open visual datasets, such as OpenConstruction, is also supporting data-driven method development. Furthermore, studies are evaluating the quality of open building datasets for mapping urban inequality, highlighting the need for high-quality, domain-specific datasets. Noteworthy papers include: The paper proposing the ConstructionSite 10k dataset, which allows researchers to train and evaluate their own VLMs with new architectures and techniques, providing a valuable benchmark for construction safety inspection. The OpenConstruction study, which synthesizes findings into an open-source catalog, supporting data-driven method development and outlining strategic priorities for future data infrastructure.