Our north star for crisis resilience✨: no one should be surprised by a natural disaster. We’ve advanced AI-based breakthrough research –  from providing timely information to forecasting and detecting natural disasters, making critical information accessible via Google products that are used by billions, and partnering with governments and organizations around the world to help communities prepare for and respond to these crises. At today’s AI for the Planet event, we shared the progress of our AI-powered tools and our continued work with global partners to help communities prepare for natural disasters: 💧Global Flood Forecasting: Our models now cover 2 billion people across more than 150 countries for severe riverine floods,  expanded from initial regional pilots to a global river flood model published in Nature. Our new Groundsource methodology now enables 24-hour advance notice for rapid-onset urban flash floods. ☂️ Predicting cyclones: WeatherNext 2 provides highly detailed hourly forecasts globally. During the 2025 hurricane season, it successfully predicted the path and intensity of cyclones with high confidence days in advance. 🔥Proactive Wildfire Detection: We have expanded wildfire boundary tracking in Search and Maps to 34 countries, including 7 new countries. And launched the first protoflight of the FireSat satellite, with future potential to detect fires just 5X5 meters anywhere on Earth with 20-minute updates when the full constellation is operational. 🌏 We’ve brought together our climate models in the Google Earth AI collection of models and datasets to enable planetary intelligence. For information to be effective, it must be actionable. Along with our research breakthroughs and product deployment, we’re collaborating with partners globally to bring helpful, timely information to people on the ground: ⚠️ We partner with authorized alert originators and distributors in over 90 countries to amplify emergency alerts and public warnings. ⚠️ Last year, Google helped connect people with crisis information over 10 million times per day, on average. ⚠️ Extreme heat alerts on Search provide warnings in over 100 countries in partnership with the Global Heat Health Information Network. ⚠️ In Nigeria and Bangladesh, GiveDirectly and the International Rescue Committee have used our flood forecasts to power anticipatory action, distributing emergency cash. ⚠️ Up-to-date air quality data is available on Google Maps in over 30 countries to help users reduce exposure to pollution. Over the past decade, we’ve made progress in climate resilience, providing actionable, timely information to communities around the world. I’m optimistic that by harnessing AI and working with our partners, we’ll move closer towards a world where no one is surprised by a natural disaster. More in today’s blog: https://lnkd.in/dq6Gp57Z

From forecasting floods to tracking wildfires, Google is building state-of-the-art AI models and global partnerships — working toward the vision: no one should be surprised by a natural disaster.🌍 Today, we are proud to share our new white paper, “How Google AI is reshaping crisis resilience,” which outlines how we’re harnessing AI to predict disasters earlier, extend warning horizons, and empower communities before crisis strikes. In 2025 alone, we helped connect people with vital crisis information over 10 million times per day on average. Here are a few highlights of how AI is making a real-world impact: 🌊 Riverine Floods: Forecasting flood events up to 7 days in advance across ~150 countries, reaching over 2 billion people. ⚡ Urban Flash Floods: Predicting localized flash floods up to 24 hours in advance using Groundsource powered by Gemini. 🌀 Cyclones: Generating up to 1,000 predictive scenarios 15 days out with WeatherNext AI, providing forecasters an extra day of critical lead time. 🔥 Wildfires: Tracking active fire perimeters across 34 countries and co-developing FireSat to detect fires as small as a shipping container (5×5m) every 20 minutes. 📱 Earthquakes: Detecting seismic waves through our global Android network to deliver early alerts within seconds. 🤝 Local Action: Partnering with groups like UN OCHA, WMO, and GiveDirectly to translate early warnings into anticipatory aid, such as direct cash transfers before floodwaters hit. There is still much more work to do, but AI is proving to be an essential tool for climate adaptation and community resilience. 📖 Read the full white paper to explore our roadmap for AI-driven crisis resilience: https://goo.gle/3Uuzu6U

Today, we are introducing Groundsource ✨, a new AI-powered methodology that transforms public information into a high-quality record of historical disaster data – starting with flash floods in urban areas. When disaster strikes, high-quality information is a lifeline. For years, our Crisis Resilience efforts at Google Research have focused on providing early warnings for natural hazards, including prediction of severe riverine flood prediction covering 2 billion people in over 150 countries. However a significant challenge remained: for many disasters, like flash floods, the high-fidelity historical data needed to train predictive AI models simply did not exist. Groundsource addresses this data gap by using Gemini to analyze decades of public information, transforming unstructured data into a structured, high-quality archive. This allows us to map the historical footprint of disasters with high precision and build a robust scientific baseline. Here is how we are applying this research to keep communities safe: ✨ Transforming Public Information:  Groundsource used news reports of 20 years in 80 languages to identify over 2.6 million historical flood events spanning more than 150 countries. ✨ Precise Mapping:  By integrating with Google Maps, the system determines exact geographic boundaries for these events, creating a dataset for flash floods. ✨ Predicting Urban Flash Floods:  Using this new dataset, we trained a model capable of predicting flash floods in urban areas up to 24 hours in advance. ✨ Expanding Flood Hub:  These new forecasts are now available on Google’s Flood Hub, complementing our existing riverine flood models that already cover 2 billion people in more than 150 countries. We are open-sourcing this dataset to provide a massive benchmark for partners and scientists to scale their impact. Groundsource joins our Google Earth AI family of geospatial models and datasets – while we are starting with flash floods in urban areas, this methodology has the potential to be applied to other hazards like landslides or heat waves, turning verified global reports into actionable datasets. By turning the records of the past into high-quality data for the future, we are moving closer to our goal: a world where no one is surprised by a natural disaster. Read about the research: https://lnkd.in/dqc9S2XY More about the Groundsource methodology: https://lnkd.in/dYshXEmS More about protecting cities with AI-driven flash flood forecasting: https://lnkd.in/dQ32uRkG Google Earth AI: ai.google/earth-ai

Since the launch of the United Nation’s Early Warnings for All initiative at COP27, Google has been supporting the UN’s mission to protect every person around the world with early warning systems by 2027. I participated in person at the inauguration as well as follow up meetings, and our teams worked in close collaboration on this important mission. The UN just released a report highlighting the critical role of AI and companies like Google in keeping communities safe. While at the Global Dialog on AI Governance and the AI for Good Summit, I discussed with many leaders the shared sense of urgency to build a safer, more resilient world. From forecasting to real-time alerting to post-disaster response, through our decade-long work here are a few examples of how we’re collaborating with the UN, governments and international organizations on global crisis resilience. 🔵 Forecasting and preparing for hazards Timely, accurate forecasts, powered by our latest AI breakthrough, enable people to take action before disaster strikes. Our forecasts are available on Flood Hub, covering 2 billion people across more than 150 countries, in areas at risk for significant flood events. In Nigeria’s Adamawa state, UN OCHA launched a Floods Anticipatory Action Programme using Google’s river flood forecasts. When forecasts indicate a high risk of significant flooding, it activates a set of early interventions such as shelter preparation. 🔵 Providing life-saving alerts in times of crisis In 2025 alone, Google helped connect people with crisis information over 10 million times per day, on average. Our Public Alerts feature data from authorities in over 90 countries so far, from partners like the US National Weather Service, the UK’s Met Office and Brazil’s Centro Nacional de Gerenciamento de Riscos e Desastres (CENAD). When authorities issue these alerts, they can appear across Search, Maps and as Android notifications. This ensures that public safety information, such as severe weather warnings and flood updates, and the practical information people need to stay safe, reaches people quickly and directly. 🔵 Using satellite imagery to accelerate disaster response Post-disaster recovery requires speed and precision. Through collaborations with UN DISHA and UNOSAT, and the use of Earth AI’s building damage assessment models, we are helping organizations analyze satellite imagery to assess infrastructure damage in hours, work that once took specialists weeks to complete. AI helped assess over 385,000 buildings following Hurricane Melissa, to accelerate disaster recovery efforts.   The full report: https://lnkd.in/gsMYmVR8 Our blog on How governments and organizations are leveraging Google’s AI breakthroughs for crisis resilience: https://lnkd.in/gCjpisQz Google Crisis Resilience site: https://lnkd.in/gNuAmbD7

This week at our Research@ flagship event, we announced new updates for our experimental Geospatial Reasoning agent and Earth AI capabilities.  Following promising tests with partners, we are expanding early access to our experimental agent to more partners.  In a new technical report, we demonstrate how combining domain-specific foundation models, spatial tools, and upfront planning enables a Geospatial Reasoning agent to carry out complex spatial analysis and planetary prediction that standalone LLMs can’t.  Highlights from this work: –> The agent connects Gemini to the capabilities of Earth AI’s foundation models  (WeatherNext, Flood Forecasting, PDFM, AlphaEarth) using modular skills, dedicated spatial workspaces, and private data support (CSV, GeoJSON). –> Beyond answering complex spatial questions, the agent powers our Planetary Prediction Engine—combining satellite embeddings and population dynamics to train validated models for disease spread, crop forecasting, or environmental risk. –> Real-World Impact with Partners:  Milken Institute: Identifying the highest-impact community resilience hubs for solar energy build out—layering rooftop yield with transit access, heat vulnerability, and local demographics data to compress 6-week siting feasibility studies into 15-min reports to unlock $179 million in clean energy capital.  WHO AFRO & DRC’s INRB: During the recent Ebola outbreak response in the DRC, teams mapped remote mining corridors by cross-referencing satellite forest loss, mining locations, and mobility data—pinpointing 48 exposed settlements and 45,500+ at-risk people in minutes rather than weeks, and modeling transmission risks to plan mobile laboratory deployment. People in Need: Analyzed historical cyclone paths to identify high-exposure villages in minutes rather than days of manual GIS work, unlocking vital time for anticipatory action. GiveDirectly: Evaluated population exposure across wind-speed zones during Cyclone Maila and benchmarked tornado damage in Mississippi within a single workspace. WPP: Connected population dynamics with proprietary data to optimize geospatial marketing while preserving privacy. –> Evaluating Geospatial Reasoning: Planetary datasets are massive and multi-layered. Across four benchmarks—including two new benchmarks we are releasing to researchers, GeoExpertQueries and GeoKTools—our evaluations quantify the essential role of specialized Earth AI tools and upfront planning over out-of-the-box LLMs, enabling the agent to execute multi-step analytical workflows reliably. Congratulations to teams across Google, and thank you to our partners for collaborating with us on the ground. Read the blog by Shravya Shetty and Luke Barrington: goo.gle/4s13F25 Early access interest form: https://lnkd.in/gxt9EfqB

Today we are announcing major advances in Google Earth AI, our collection of geospatial AI models and reasoning. For years, Google’s geospatial AI models have provided measurable, life-saving impact: 💧 Flood forecasting models now covering over 2 billion people in 150 countries for severe riverine floods. 🔥 Providing wildfire boundaries in Google Maps for 27 countries. 🌀 Improving accuracy for tropical cyclone prediction. 🚗 Reducing traffic stops and emissions in 20 cities across four continents. 🔎 Specialized AI models enable finding objects, classifying scenes or monitoring changes across Earth’s observation imagery 🧑🤝🧑 Insights on the dynamics of human behavior across the planet 🚍 AI advancements in measurement, simulation, and optimization to provide transportation agencies with powerful tools for management of urban transportation systems. In our new technical paper, we introduce new Remote Sensing Foundations and Population Dynamics Foundations models and demonstrate the power of our geospatial reasoning agent. 🌍 Geospatial Reasoning: answer complex, multi-modal questions with the help of Gemini-powered agents that orchestrate Earth AI models. Now an analyst can, for example, not just see where a storm might hit but also identify vulnerable communities and at-risk infrastructure. 🌍 Deeper Insights in Google Earth: we are adding Earth AI models to Gemini capabilities in Google Earth, letting users search for objects in satellite imagery using natural language. 🌍 Expanding Access via Google Cloud: we are making the core Earth AI Imagery, Population, and Environment models directly available to trusted testers on Google Cloud. We are committed to translating AI breakthroughs into systems that address large-scale, real-world problems. This work represents a critical step in building AI that can reason about the physical world with the same fluency that Large Language Models reason about text. Congratulations to the dedicated teams across Google Research, Google Maps Platform, Google DeepMind, and Google Cloud who are bringing this vision to life. Read our announcement with Christopher Phillips: https://lnkd.in/dDs7XEpM Research blog by Niv Efron and Luke Barrington: https://goo.gle/4oyoA9O Technical paper: https://lnkd.in/dg28pQUZ More on Google Earth AI: https://ai.google/earth-ai Sign up to be considered for a Trusted Tester for Geospatial Reasoning: https://lnkd.in/d34aeXny

We are taking the next step in our flood resilience work. Over several years, Google Research has built state-of-the-art AI models for more accurate flood forecasting, ensuring this technology reaches frontline responders to give them critical time to act. Based on these models, we show in our flood hub forecasting in 150 countries, covering 2B people for severe riverine floods. We are now open-sourcing our hydrology modeling framework on GitHub. We believe that a scientific breakthrough reaches its full potential when it empowers others to replicate and expand upon findings, ensuring that innovation is a catalyst for worldwide progress. Key aspects of this release: 👉 Empowering the Community: Researchers and operational forecasters can now access the river forecast model’s architecture and training pipeline that is used for Google’s Flood Hub. This allows local experts to train models using their own specialized data, retaining full control while benefiting from state-of-the-art AI-based flood forecasting. 👉 Technical Progression: This repository includes two distinct versions: our original 2024 benchmarked model and an upgraded v2 model. The v2 model utilizes an ME-LSTM architecture to process diverse meteorological inputs into a unified flood prediction system, extending our reliable predictive horizon by six days in gauged basins and one day in ungauged basins. 👉 Operational Integration: We collaborated with the Czech Hydrometeorological Institute (CHMI) to validate that our AI-based model provides forecasts comparable to traditional, locally calibrated conceptual models. CHMI also developed an adapter for the Delft-FEWS platform, creating a blueprint for global agencies to enhance their water management workflows with machine learning. 👉 Expanding Access: By providing readily adoptable open-source tools, we aim to bridge the gap between technological innovation and real-world effectiveness, particularly in resource-constrained regions that lack costly traditional forecasting infrastructure. This framework is now live on GitHub under an Apache 2.0 license: https://lnkd.in/gDzC6GkF Read more about our hydrology framework and how to get started on the Google Research blog:  goo.gle/4x9p2R3 On our approach to open science: https://lnkd.in/gjnCvkTK

Today, Google DeepMind and Google Research are introducing WeatherNext 3 ✨ Our most advanced and accurate AI weather model now learns directly from real-time observations, enabling it to provide timely and more localized predictions for the weather events that impact people the most. WeatherNext 3 now includes real-time satellite data, hourly refreshes, higher resolution, precise precipitation forecasting, and clean energy variables. Here are the key aspects of this release: → The biggest leap forward is that the model learns from global geostationary satellite data and thus bypasses the typical six-hour data lag of traditional numerical weather prediction models. This allows the model to generate a new forecast every hour, each one grounded in the most recent satellite observations available, at up to 5-kilometer resolution. This is particularly important for critical weather that develops fast. → Precipitation accuracy improvement is quite significant as well: we train our model on two exceptionally high-quality sources of precipitation data: NASA’s satellite-based Integrated Multi-satellite Retrievals for GPM (IMERG) and our own global precipitation reanalysis based on satellite radar.  As a result, WeatherNext 3 achieves a Continuous Ranked Probability Score (CRPS) improvement of up to 60% against IMERG. → Higher Resolution & Frequency: WeatherNext 3 generates hourly forecasts at multiple spatial resolutions, providing a global picture roughly five times sharper than its predecessor. → Clean Energy Variables: The model introduces specific forecasts for 100-meter wind speeds, cloud cover, and sun radiation to assist global clean energy planning and grid operations. WeatherNext 3 is rolling out globally across Search, the Gemini app, Google Maps, Google Maps Platform, and Cloud to improve everyday planning and long-term forecasts. As we continue to advance geospatial foundation models, WeatherNext 3 brings us closer to a future where forecasts truly match what is happening on the ground. Announcement blog: https://lnkd.in/dkTE267E Build with WeatherNext 3: https://lnkd.in/dtJtsRvP Learn more about Google Earth AI for planetary intelligence: https://lnkd.in/d33UYtae

Today, Google is releasing our 11th Environmental Report, providing a comprehensive view of our 2025 sustainability performance. In 2025, our AI infrastructure buildout accelerated, placing new demands on energy grids globally. While our operational growth was significant, we remain focused on decoupling that growth from our carbon footprint through a full-stack approach to AI. Key highlights from the report: 👉 Record Clean Energy: Google signed agreements for over 12 GW of net-new clean energy in 2025. Cumulatively, we’ve contracted 35 GW of clean energy since 2010, making Google one of the largest corporate purchasers of clean energy in the world. 👉 Decoupling Growth from Emissions: Despite a 37% annual increase in electricity demand, we reduced operational emissions by 2% year-over-year. 👉 Efficiency at Scale: Google’s data centers use 83% less overhead energy than the industry average. Through hardware, software, and compute efficiency improvements, we avoided over 58 million metric tons of CO2e in 2025—our carbon footprint would have been five times larger without these interventions. And we keep driving research and innovation for AI efficiency. We are actively applying foundational research to some of the world’s most pressing environmental challenges. AI is increasingly serving as a catalyst for mitigation, resilience, and nature protection. 🌎 AI as a Force Multiplier for Mitigation: In 2025, nine of our solutions collectively enabled partners to reduce an estimated 41 million tCO2e—roughly three times our own emissions. Solutions like Contrails, Solar API, and Green Light have tremendous runway to make impact. 🌎 Geospatial Reasoning for Public Health & Planning: We are evolving Google Earth AI to synthesize diverse geospatial datasets towards planetary intelligence, helping partners with proactive insights for resource allocation and health emergency preparedness. 🌎 Crisis Resilience: Our AI-powered flood forecasting now provides up to seven-day predictions for over 2 billion people across more than 150 countries. Additionally, our FireSat satellite constellation in collaboration with the Earth Fire Alliance and Muon Space aims to spot wildfires as small as a garage. More in my recent post on Crisis Resilience. 🌎 Bioacoustics for Biodiversity: Tools like Perch 2.0 and SpeciesNet are helping the research community scale insights into marine and terrestrial ecosystems, allowing us to monitor biodiversity with greater precision than ever before. Reaching our climate and energy ambitions requires significant, continued innovation. While the path is non-linear, we remain dedicated to a strategy of steady execution, balancing our bold moonshots with real-world impact. Read the full report: https://lnkd.in/eF8AcYPA Blog by Kate Brandt: https://lnkd.in/ey-DNd6e

It was only a bit over a year ago that we watched the launch of the first FireSat protoflight, marking a major milestone in efforts to modernize global wildfire detection. 🚀 Today, I am pleased to share that three new FireSat satellites have successfully launched from Vandenberg Space Force Base in California, expanding the FireSat program. This global initiative, led by the Earth Fire Alliance  and built by satellite manufacturer Muon Space, aims to establish an unprecedented dataset for wildfire detection and monitoring. Google Research has collaborated with these partners to develop a purpose-built constellation designed to provide the continuous coverage necessary to detect wildfires before they spread. Key aspects of this mission include: 🛰️ Building on Proven Technology: This launch follows the successful pilot mission last year, which validated the specialized sensor architecture capable of detecting early-stage wildfires as small as 5×5 meters. These initial tests confirmed the constellation’s ability to spot small, low-intensity blazes that frequently remain invisible to existing satellite systems. 🛰️  From Pilot to Constellation: This progression is essential for providing the global, near-real-time situational awareness that fire agencies require to mitigate the impact of increasingly intense fire seasons. 🛰️  Cross-Sector Collaboration: The initiative is a direct result of diverse entities—technology companies, nonprofits, philanthropy, and the private sector—uniting under a shared mission. With funding from Google.org, Bezos Earth Fund, Gordon and Betty Moore Foundation, and others FireSat took another tangible step forward in putting practical AI to work for climate resilience. This launch is another tangible step towards our north star for crisis resilience efforts: for no one to be surprised by a natural disaster coming their way. For further reading on the FireSat initiative and our ongoing work in wildfire detection: https://lnkd.in/gnnCgaMg https://lnkd.in/gGFzwYNG On today’s launch: https://lnkd.in/gDdde8Hg Picture credit SpaceX