Google Research has expanded its Heat Resilience dataset to cover more than 50 global cities, releasing building-level rooftop reflectivity data designed to help urban planners tackle the urban heat island effect. The initiative leverages machine learning to fuse high-resolution satellite imagery (30-cm) with lower-resolution global satellite data to map individual building surfaces with unprecedented precision. The expanded dataset is now accessible through a new Heat Resilience Earth Engine App, empowering cities worldwide to identify neighborhoods most vulnerable to extreme heat and prioritize cool-roof interventions. The research, published in Nature Communications, demonstrates that targeted cool-roof planning using this data could mitigate extreme urban heat by up to 0.5°C (0.9°F) globally. The AI-driven methodology combines Sentinel-2 satellite data with commercial imagery from Airbus Pléiades Neo, achieving a root mean square error of just 0.04 relative to ground-truth hyperspectral measurements. The approach addresses a critical gap in urban climate planning, where neighborhood-level averages typically obscure the individual buildings most suitable for reflectivity retrofits. The expanded dataset covers over 50 cities across 9 countries, including major urban centers in Europe, Brazil, and the United States. Google's pilot program in 2024 with 14 cities already yielded tangible results, guiding the implementation of cool-roof ordinances and municipal adaptation plans. With approximately 500,000 deaths annually attributed to extreme heat, Google's open-access platform represents a significant tool for municipal governments seeking cost-effective solutions to urban temperature elevation.