Impact Signals #104: Federal Child-Welfare Predictive Analytics, 15,000 Meta Glasses for Vision Ireland, AI Robocalls Close Georgia Schools The Administration for Children and Families is sending $6 million to ten jurisdictions to put predictive analytics inside decisions about whether a child is removed from a home, with three-year pilots starting September 30. Meta is donating 15,000 Ray-Ban glasses to Vision Ireland, enough for every blind adult the charity supports, while an AI-voiced robocall closed a Georgia school district for a full day and left no suspect and no device behind. Elsewhere, Nigeria certified 450 teachers to teach AI literacy in three displacement-heavy states, and Khyber Pakhtunkhwa connected 175 public schools to an AI teacher that answers in Pashto. Today the story is not who wrote a check, it is who is being handed an algorithm. --- Top Stories **1. Administration for Children and Families Puts $6 Million Into Predictive Analytics Across Ten Child-Welfare Jurisdictions** The Administration for Children and Families, the HHS agency that oversees child welfare, announced on August 13 that it is sending $6 million to ten jurisdictions to build and test predictive analytics inside their child welfare systems, roughly $600,000 each. The recipients are Indiana, Missouri, Nebraska, New Jersey, North Carolina, Ohio, Oklahoma, Texas, the District of Columbia and the Muscogee Creek Nation, and the three-year pilots begin September 30, 2026. ACF names the inte…

Impact Signals #103: AIDE Counts 1,100 Emergency-Management AI Products, OpenAI Foundation Stakes $100M on Hepatitis C Cure Rates, Rural Patients Push Back The Markle Foundation's AIDE Initiative published the first national assessment of AI in US emergency management on August 11, counting more than 1,100 products from over 700 companies against local offices that in many counties run on a single full-time staffer. Two days later the Common Health Coalition announced $100 million from the OpenAI Foundation to at least double hepatitis C cure rates in eight states and localities inside two years, starting with Alabama, Illinois, Louisiana and Massachusetts. On the same day KFF Health News reported from Hot Springs, South Dakota that the entire published evidence base for AI in rural health care runs to 26 peer-reviewed studies since 2010, while states steer money from a $50 billion federal program toward it. Today the money and the mandates arrive together, and the binding constraint everywhere is people, evidence and bandwidth rather than software. --- Top Stories **1. AIDE Initiative Counts 1,100 Emergency-Management AI Products, Then Names Staffing as the Binding Constraint** The AI for Disasters and Emergencies (AIDE) Initiative, led by the Markle Foundation with Aspen Digital and RAND, published its assessment of AI across the US emergency management community on August 11, 2026. It ships as three documents: the main report "AI for Disasters and Emergencies: A Way Forwa…

Impact Signals #102: Gates Foundation's $540M AI Health-Data Bet, GiveDirectly's Three-Week Earthquake Cash, an AI Supercomputer for East Africa's Climate Center The Gates Foundation just made the largest charitable gift in University of Washington history: $540.2 million over ten years to the Institute for Health Metrics and Evaluation, expanding the Global Burden of Disease study from 925 to nearly 5,000 locations and funding AI-enabled tools inside the dataset the world's health planners steer by. On the operational side, Fast Company documented GiveDirectly landing cash on Venezuelan earthquake survivors' phones three weeks after the shock, guided by an AI targeting chain that people verified rather than drew. In Nairobi, the World Bank is financing a GPU server for the IGAD climate center, and Philippine agriculture officials are moving drone-and-AI crop surveillance from pilot to policy. Today's signal is money meeting infrastructure: private and development capital paying for the data backbones, compute, and targeting systems that public budgets are pulling away from. --- Top Stories **1. Gates Foundation Commits $540.2 Million to IHME to Build AI-Enabled Health Data as US Funding Retreats** On August 10 the Gates Foundation announced a $540.2 million, ten-year grant to the Institute for Health Metrics and Evaluation (IHME) at the University of Washington, the largest charitable gift in the university's history. The money expands the Global Burden of Disease study fro…

Impact Signals #101: Canada Redraws Its Storm Warnings, India Builds a National AI Early-Warning Platform, Code.org Becomes CodeAI Environment and Climate Change Canada switched on free-form polygon alerts for severe thunderstorms and tornadoes, warning only the people a storm is actually expected to hit. In India, the government told Parliament it is building a national AI-powered early-warning platform on 50 Doppler radars and nearly 7,000 rain gauges, while in Nairobi two schoolgirls from Wajir won the national science fair with an AI drought predictor built for the communities that live under the hazard. Today's signal: warning systems are getting rebuilt at national scale, and Colombia's strongest earthquake this century is testing every one of them in real time. --- Top Stories **1. Canada Redraws Its Storm Warnings: Free-Form Polygons Replace Zone Blasts to Fight Alert Fatigue** Environment and Climate Change Canada launched its Convective Alert Modernization system on Tuesday, August 11. Instead of warning entire pre-drawn forecast zones, meteorologists now draw a free-form polygon around the area a severe thunderstorm or tornado is actually expected to hit, and only people inside that shape get the warning. Forecasters can track a storm in real time, splitting or merging warning areas and upgrading one portion of a storm without re-alerting everyone else. The change covers severe thunderstorm and tornado warnings; heat warnings and thunderstorm watches stay on the b…

Impact Signals #100: Oregon Fields Mobile AI Wildfire Cameras, Korea's Marine Heatwave Kills 860,000 Farmed Fish, AVPN Opens an AI Fellowship for Nonprofit Leaders The Oregon Hazards Lab is trucking solar-powered AI wildfire cameras to active fires in Eastern Oregon, closing in hours the coverage gaps a fixed tower network leaves open for years. Off the Korean coast, a record-hot ocean has killed more than 860,000 farmed fish, the exact event class the country's new 38-hour AI marine-heatwave warnings were built to get ahead of. And in Fortune, two nonprofit-sector veterans tell AI's new millionaires that the $600 billion US charitable economy needs their funding, not their reinvention. Today's signal: detection infrastructure is going mobile while the climate outruns the warnings. --- Top Stories **1. Oregon Hazards Lab Fields Mobile AI Wildfire Cameras That Deploy to Active Fires Within Hours** The Oregon Hazards Lab at the University of Oregon has begun fielding the Deployable Universal Camera Kit, a mobile wildfire-detection unit built around a pan-tilt-zoom camera, a lightweight mast tower, a solar-charged battery pack, and a Starlink uplink, designed to stand up within hours in places the fixed camera network cannot see. Four mobile units are deployed now, including at active fires in Eastern Oregon. The mobile feeds plug directly into ALERTWest, the AI-monitored western wildfire camera network KEZI reports is now the largest of its kind, so temporary units get the sam…

Impact Signals #99: World Bank's First AI Flagship Report, New Orleans Corrects the 911 AI Story, Xcel Files a Wildfire AI Plan With Texas Regulators The World Bank has put its flagship research product behind AI for development for the first time: World Development Report 2026 landed August 4, and this morning's India coverage adds field numbers, including net savings of up to $560 per farmer from a Telangana forecasting pilot and 2.7 million chatbot queries answered in 186 days. In New Orleans, the Orleans Parish Communication District went on the record to correct national headlines about AI answering 911 calls, publishing the actual activation rules. And Xcel Energy filed a Texas Panhandle wildfire plan built on AI detection cameras with the state's utility regulator. Today's signal: AI for impact is moving out of press releases and into records, a flagship report, a regulatory docket, an on-the-record correction. --- Top Stories **1. World Bank's First AI Flagship Report Says Developing Economies Could Compress a Century of Progress Into a Decade, With India as Proving Ground** The World Bank released World Development Report 2026, "The Promise of Artificial Intelligence," on August 4, its first comprehensive assessment of what AI means for developing economies. Its headline claim: AI could let developing countries do in a decade what might otherwise take a century, if governments close gaps in power, connectivity, skills, and institutions. It also lands a counterintuit…

Impact Signals #98: Gujarat Flags 167,000 At-Risk Students, Ahmedabad's AI Command Centre Counts Lives, WHO's Climate-Migration Health Agenda Gujarat's Education Department, working with Wadhwani AI and UNICEF India, is using an AI early-warning system inside the state's own school census to flag students likely to drop out, 167,000 in its first year and another 118,000 for 2026-27. In Ahmedabad, Marengo CIMS Hospital published a two-and-a-half-year tally from its 24/7 AI monitoring command centre: 836 deteriorations caught early and 115 code blue emergencies avoided across 6,551 high-risk patients. And the World Health Organization released its first global research agenda on health, migration and climate change, the document funders will cite when they decide which studies at that intersection get money. Today's signal: AI is moving into the institutions that already hold the data (a school census, a cardiac ward, a health system's follow-up calls), and the results are being counted in students kept and emergencies avoided, not in demos. --- Top Stories **1. Gujarat's Dropout Early-Warning System Flags 167,000 At-Risk Students Before They Disappear** Gujarat's Education Department, with Wadhwani AI and UNICEF India, has built an AI early-warning system into the state's existing Child Tracking System that predicts which primary-school students (Classes 1 to 8) are likely to drop out, using data the state already collects: attendance, assessment scores, age, gender, disabili…

Impact Signals #97: DeepMind Open-Sources a Cyclone Model That Gains a Full Day of Warning, Bangladesh Closes the Last-Mile Language Gap, US School AI Surveillance Loses Its Federal Check Google DeepMind published peer-reviewed results in Nature showing its WeatherNext cyclone model hands forecasters an extra day of accurate warning, then open-sourced the model so met agencies without supercomputers can run it themselves. In Bangladesh's Chittagong Hill Tracts, volunteers translating national disaster alerts into local languages got usable warnings to communities before July's landslides, and in the US, the Department of Education's removal of disparate-impact enforcement has left school AI monitoring, which survey data says has outed nearly a third of students, without its main federal check. Today's signal is the widening gap between what early-warning AI can now do and the institutional plumbing, language, governance, and accountability that decides who actually benefits. --- Top Stories **1. DeepMind's Cyclone Model Gains a Full Day of Warning, and Google Is Giving It Away** Google DeepMind published peer-reviewed results in Nature on August 6 showing its WeatherNext cyclone model predicts tropical cyclone track, intensity, and wind structure at state-of-the-art accuracy: its three-day forecasts match what the best physics-based models previously achieved only at two days, effectively an extra day of lead time. The model was trained on roughly 20 terabytes of reanalysis…

Impact Signals #96: New Orleans 911 AI Disclosure, Data Centers Near 1,900 US Schools, Korea's 21-Hospital AI Pilot New Orleans confirmed on August 6 that an AI built with Carbyne has been answering certain 911 calls since 2023, and callers are never told — the disclosure came only after a viral post forced the issue. The same day, Brown University researchers counted 1,878 US public schools within a mile of a data center and linked proximity to lower math scores, while Seoul National University Bundang Hospital kicked off a 21-institution national pilot wiring 19 approved AI tools across the full patient journey. Today's signal: AI systems already embedded in emergency lines, school zones, and hospital wards are now being counted, confirmed, and formalized — the accounting is catching up to the deployment. --- Top Stories **1. New Orleans Confirms an AI Has Been Answering Some 911 Calls Since 2023 — Callers Are Never Told** The Orleans Parish Communications District confirmed on August 6 that an AI system built with emergency-call software firm Carbyne has been answering certain 911 calls since 2023. The tool triggers only when a caller dials from within 200 meters of a car accident already logged in dispatch: it confirms the caller is reporting the same crash, says the incident is handled, and asks them to stay on the line only if they are directly involved or have new information. OPCD says the AI now absorbs 15 to 18 of every 20 calls about a single accident, freeing rou…

Impact Signals #95: Telangana's AI Emergency Line Goes Live, China's Five-Year Disaster Plan, EU AI Act Enforcement Begins Telangana switched on an AI-driven statewide 112 emergency line on August 5 and published its before-and-after numbers the same day: call-centre processing time down 30 percent, average response now 10 minutes 30 seconds against India's 18-minute national average. China's Ministry of Natural Resources released a five-year geological disaster plan that writes large AI models into national policy while relocating 50,000 households out of high-risk zones, and the EU AI Act's high-risk obligations became enforceable on August 2, making accessibility a legal market-access condition across 450 million people. Today's signal is AI for public services crossing from announcement to accountability — live systems with published metrics, plans with numeric targets, and laws with enforcement teeth. --- Top Stories **1. Telangana Turns On an AI-Run 112 Emergency Line, Cutting Average Response Time to 10 Minutes 30 Seconds** On August 5, Telangana's Director General of Police CV Anand inaugurated the upgraded TG-ERSS Dial 112 system at the Telangana Integrated Command and Control Centre in Hyderabad. The platform folds police, fire, ambulance, women's safety, child protection, and disaster response into the single national emergency number 112, using AI to filter duplicate and non-emergency calls, prioritize incidents by severity, and route calls to the right service.…