#12: Philippines NAICRI, India SAHI, Cyclone Horacio, Vanuatu Recovery
AI for Impact Daily Briefing — February 24, 2026 (Tools & Platforms)
🔥 Top Stories
1. Philippines Launches NAICRI: National AI Hub with Shared Disaster Model Repository
The Department of Science and Technology formally launches the National Artificial Intelligence Center for Research and Innovation (NAICRI) on February 26. Three operational components: DIMER (Democratized Intelligent Model Exchange Repository) — a national commons of pre-built, field-tested AI models for disaster detection, agriculture, and traffic analysis; NAIRA (AI-as-a-Service hub) — lets government agencies and NGOs access tools without ML engineers on staff; and iTANONG, a natural language query tool for government data. DIMER directly targets the sector's most persistent barrier: the "build from scratch" problem that causes every new typhoon response to reinvent damage assessment tools from zero. Philippines ranks top-5 globally in disaster exposure. If DIMER works here, it's a transferable template for every disaster-prone country in the region.
2. India's SAHI Framework: AI Health Governance at Population Scale
India's Ministry of Health released SAHI (Strategy for Artificial Intelligence in Healthcare) alongside BODH — a benchmarking platform for health AI tools. Unlike aspirational policy documents common across lower-income nations, SAHI governs infrastructure already running at population scale: eSanjeevani telemedicine has logged 282 million consultations with AI-generated differential diagnosis recommendations; Ayushman Bharat Digital Mission has national health IDs and interoperable records deployed; AI media scanning has published 4,500+ infectious disease event alerts to district governments. BODH provides standardized testing and scoring before deployment — reducing procurement risk for humanitarian health NGOs. SAHI's third governing principle is a deliberate break from WHO and EU precautionary models: "responsible innovation should be prioritised over cautionary restraint." This framing will shape how other countries in the Global South approach AI health governance through 2026.
3. Cyclone Horacio: First Category 5 of 2026 — Live Test for AI Forecasting
Tropical Cyclone Horacio intensified into the first Category 5 storm of 2026 on February 24 — sustained winds of 161 mph, tracking southwest of Rodrigues Island in the Indian Ocean. JTWC is running AI ensemble models alongside traditional ECMWF and GFS forecasting, and DeepMind's weather AI is in the mix. The central question being tracked: can AI catch rapid intensification events — a storm gaining 35+ mph in 24 hours — earlier than traditional numerical weather prediction? This has historically been the hardest problem in tropical meteorology. PDC Global is providing real-time dashboards combining AI analysis with satellite data for disaster managers in Madagascar, Mozambique, and the Mascarene islands. Post-season analysis will benchmark AI model performance specifically for rapid intensification events, making Horacio a documented test case.
4. Vanuatu: Japan + UNICEF Launch Disaster-Resilient Health Programme
On February 10, the Government of Vanuatu — with Japan bilateral funding and UNICEF technical support — launched "Strengthening Disaster Resilience of Health Care Facilities in Remote Islands." The programme will upgrade 20 primary health facilities to cyclone-resistant standards, serving 30,000+ people. Baseline conditions illustrate the gap: 25% of facilities have no reliable electricity (no vaccine cold chain, no equipment at night); 15% have no water access; 89% lack essential equipment for mothers and newborns. Vanuatu has experienced multiple Category 5 cyclones and a 7.4 magnitude earthquake since 2023, the latter disrupting services for 80,000 people. Vanuatu is simultaneously managing an active pertussis outbreak with 7 deaths — vaccination campaign ongoing. The Japan bilateral + UNICEF delivery model is directly transferable to Fiji, Tonga, Samoa, and other Pacific small island developing states facing identical vulnerabilities.
5. HLA Pulse Survey: 1,729 Humanitarians Rate AI Adoption Barriers
The Humanitarian Leadership Academy surveyed 1,729 humanitarian workers across 120+ countries in January 2026 — the broadest practitioner-level AI assessment in the sector to date. Primary finding: high interest in AI tools, but training gaps and organizational policy uncertainty are the two biggest brakes. This captures practitioners in under-resourced contexts, not only well-funded Northern NGOs. On February 26 — this Thursday — HLA hosts a free virtual webinar presenting the full results. If your organization is building a case for AI investment, skills budget, or digital transformation planning: this data is what you bring to that conversation. Register at humanitarianleadershipacademy.org.
6. Airbnb.org + Mexico City: AI-Enabled Emergency Shelter Goes Formal
Airbnb.org and Mexico City's Ministry of Integrated Risk Management (SGIRPC) signed a formal 3-year collaboration agreement February 12, integrating residential housing inventory into emergency response infrastructure before disaster strikes. The key innovation: no procurement scramble when a hazard activates — the protocol is pre-negotiated. AI demand forecasting identifies which neighborhoods need surge housing capacity based on hazard modeling, pre-positioning supply. Built on an October 2025 pilot during Veracruz floods. Mexico City has 22 million residents on a former lake bed; the 1985 and 2017 earthquakes each caused large-scale displacement. The Mexico City model — formal pre-disaster housing protocol combined with AI-assisted demand forecasting — is replicable for Manila, Lima, Dhaka, and any dense urban area in a seismic or flood zone.
7. UNDRR + ECOWAS: Copernicus Satellite Tools for West Africa — The Last-Mile Problem
UNDRR and ECOWAS published findings from a December 2025 regional webinar drawing 170 participants from 39 countries. The European Commission's Joint Research Centre presented Copernicus Emergency Management Service — EU satellite constellation providing real-time monitoring of floods, wildfires, and droughts — as the cross-border early warning tool for West Africa. Critical finding surfaced: Copernicus is only as good as the ground-truth data fed back into it. If national agencies don't consistently upload ground-level measurements, satellite projections lose local accuracy. The "last-mile data gap" remains the bottleneck across all earth observation tools in low-resource settings. Any organization using Copernicus in West Africa should be inside the ECOWAS data-sharing network — contributing ground data and receiving improved projections — not only consuming the satellite feed.
8. The "AI Slop" Crisis Threatening Humanitarian Open Source
A systemic crisis is peaking in open-source software with direct implications for humanitarian tooling. AI-generated code submissions — often from automated agents without human review — are flooding maintainers with invalid pull requests and bug reports. cURL, which powers a significant share of internet infrastructure including humanitarian data systems, shut down its bug bounty program after receiving 20 invalid AI-generated security reports in 21 days. Ghostty implemented a permanent ban on AI-generated submissions. Open-source tools that underpin humanitarian work — OpenStreetMap processors, Humanitarian Data Exchange utilities, crisis NLP tools — rely on volunteer maintainers. A maintainer spending 80% of review time rejecting AI spam has 20% left for legitimate contributions. Two immediate actions: require rigorous human review before any AI-assisted code submission to critical infrastructure; consider financially supporting open-source humanitarian tool maintainers via GitHub Sponsors or direct grants.
📅 Upcoming Events
HLA AI Pulse Survey Results Webinar
- Date: February 26, 2026 (This Thursday)
- Location: Virtual (free)
- Why it matters: Primary data from 1,729 humanitarian practitioners across 120+ countries on AI adoption barriers — essential for any organization building an AI investment case.
- Link: Register
Philippines NAICRI Official Launch
- Date: February 26, 2026
- Location: Manila
- Why it matters: National AI hub launches with DIMER disaster model repository — field-ready AI tools for disaster response practitioners in Southeast Asia.
- Link: DOST
Humanitarian Networks & Partnerships Week (HNPW 2026)
- Dates: March 2–6 (Virtual), March 10–12 (Hybrid Geneva)
- Location: Geneva, Switzerland / Virtual
- Why it matters: OCHA's flagship annual humanitarian coordination gathering. Anticipatory action, AI/data tools, and early warning all on the agenda.
- Link: Register
Summit of AI in Latin America (SALA 2026)
- Dates: March 9–12, 2026
- Location: Quito, Ecuador
- Why it matters: Premier regional AI event with hackathon on humanitarian challenges; government, academia, and civil society from across Latin America.
NVIDIA GTC 2026
- Dates: March 16–19, 2026
- Location: San Jose, California (Hybrid)
- Why it matters: Earth-2 climate AI sessions; DLI training directly applicable to satellite imagery and crisis informatics.
- Link: Register
🌍 Active Disaster Monitoring (GDACS/OCHA)
- Cyclone Horacio (Indian Ocean): 🔴 Active — Category 5, 161 mph winds as of Feb 24. Tracking southwest of Rodrigues Island. PDC Global dashboards active for Madagascar, Mozambique, Mascarene islands.
- Vanuatu: 🟡 Active — Pertussis (whooping cough) outbreak, 7 deaths. Ministry of Health reports declining transmission; vaccination campaigns ongoing.
- Colombia: 🔴 Active — Floods since January 26. 72,000 families affected, 44 deaths, 12,000 homes damaged. Government emergency declared.
- Malaysia/Indonesia: 🟡 Active — Flooding across Central Java and Peninsular Malaysia from ASEAN Week 7 events.
- Sudan: 🔴 Ongoing Crisis — WFP digital beneficiary management active amid humanitarian access constraints and convoy disruptions.
Sources: PNA.gov.ph, PIA.gov.ph, ICTworks.org, PIB India, AnalyticsInsight, JTWC, PDC Global, UNICEF Pacific, RNZ News, Humanitarian Leadership Academy, Mexico Business News, ReliefWeb, UNDRR, InfoQ, RedMonk, Gemini Deep Research