On a sweltering June morning in Pittsburgh, as wildfire smoke drifted east from thousands of miles away, researchers at Carnegie Mellon University flipped on their screens to begin testing a new kind of emergency dashboard—one that could show not just where people are, but where they’re fleeing to, in real time. This isn’t science fiction: it’s the result of a groundbreaking partnership between Carnegie Mellon’s NSF AI Institute for Societal Decision Making and Meta’s AI for Good program, launching in 2026 to transform how emergency responders manage disasters. The collaboration aims to turn vast streams of anonymized, aggregated mobility and connectivity data—collected through Meta’s platforms—into intuitive, AI-powered visualizations that first responders can use during hurricanes, wildfires, and severe winter storms. In moments when every second counts, these dynamic situation reports could mean the difference between life and death.
Natural disasters are growing more frequent and intense, and emergency teams are often overwhelmed by fragmented information. Traditional methods of tracking population movement—like surveys or traffic cameras—lag behind real-time needs. This new system changes that. By combining Meta’s open-source AI models, including Segment Anything and DINO, with satellite imagery and large language models, the team is building tools that can detect evacuation compliance, identify stranded communities, and even spot when residents begin returning after a storm has passed. The data is anonymized and aggregated, ensuring privacy while still offering granular insights. For Rebecca Nugent, head of CMU’s Department of Statistics and Data Science, the project embodies the power of collaboration: “This NSF AI-SDM-Meta collaboration is an excellent example of how academic-industry partnerships can impact social good.”
The tools will be evaluated during actual disasters in 2026, giving researchers the chance to refine them under real pressure. Aarti Singh, director of the NSF AI-SDM and a professor in CMU’s Machine Learning Department, emphasized the deeper goal: “Our partnership with Meta solidifies an important informational piece relevant to AI-SDM’s effort on designing effective disaster risk communication by understanding human mobility and networking behavior.” By integrating social science with cutting-edge AI, the team isn’t just tracking movement—they’re decoding human behavior in crisis. Early applications could include alert systems that adapt messaging based on where people are actually going, not where officials assume they’ll go.
If successful, these tools could be shared globally, offering a blueprint for disaster response in vulnerable regions from Southeast Asia to sub-Saharan Africa. In a world where climate shocks are the new normal, the ability to see and respond to human movement with clarity and speed may become one of our most vital defenses. This summer, as storms form and temperatures rise, a quiet revolution in emergency response is unfolding—one algorithm, one dashboard, one life at a time.