#122: Impact Signals #122 — Colombia's Volunteer AI Aid Map, Philippines Mandates School Hazard AI, Chicago Tops Climate Ranking
AI for Impact Daily Briefing, September 05, 2026
Top Stories
After Colombia's Earthquake, Volunteers Built an AI Aid Map Before the Government Did
A magnitude-7.4 earthquake knocked out water, power and cell signal for more than 1.5 million people in Colombia. Civilians opened ad hoc donation points in homes, cafes, churches, gyms and a brewery, but supply matching turned chaotic — sites with food had no water, sites with water had no clothes, and on one day everyone needed underwear. Tech volunteers built a site, first called Ayudas Pereira and later renamed alluda.online, letting drop-off points list what they had and letting people request or offer goods and transport, with AI used to organize incoming reports, spot patterns, and flag communities that hadn't been reached yet. Three days after the quake the site had 100 logged donation points across 25 cities and towns, though it briefly crashed after 23,000 visits and has no staff to verify content or block fraud. Why it matters: The tool that closed Colombia's coordination gap wasn't built by a UN agency or a preparedness-funded NGO — it was assembled by tech workers within hours of the quake, using off-the-shelf AI to do triage a spreadsheet couldn't keep up with. The same speed produced its weak point: alluda.online still has no verification layer or anti-fraud check, and it buckled once at 23,000 visits — a gap that remains open as the site keeps running unstaffed.
The Philippines Just Made an AI Hazard Tool Mandatory for Every Public School
Education Secretary Sonny Angara has directed public school administrators, teachers and disaster coordinators nationwide to fold LIGTAS+ (Learning Institution Geohazard Tracking and Assessment for Safety) into their disaster routines, as hundreds of schools remain affected by flooding. The tool combines AI with weather forecasts, satellite radar imagery and government/international scientific data, running a four-stage decision framework — continue, ease-in, check-in, stop — to guide whether a school keeps class, shifts to remote learning, or suspends outright. DepEd says a report that used to take hours to compile manually during a typhoon can now be generated in about five minutes. Why it matters: This converts a discretionary tool into a national mandate for the exact decision — cancel class or not — that flood-prone school systems most often get wrong or get too late, and DepEd is willing to put a number on the time saved (hours to five minutes) rather than just claim speed. LIGTAS+ is now the default decision path for every public school administrator in the country.
A Geospatial-AI Ranking of 72 Cities Put Chicago on Top and Ahmedabad at the Bottom
AlphaGeo, a geospatial-AI company that models and predicts location-level climate risk, ranked 72 of the world's largest cities on climate-disaster preparedness for the New York Times. Chicago came out on top; Ahmedabad, India, ranked last. Within the U.S., Phoenix was named the highest-risk city as its heat has intensified through 2026. The ranking is built to separate cities that have invested in resilient infrastructure from those that haven't, at a moment when that gap is widening. Why it matters: A single, comparable AI-generated risk score across 72 cities gives planners, insurers and NGOs a common reference point for where to prioritize resilience spending, instead of relying on each city's own self-assessment. AlphaGeo's full 72-city list is now the reference ranking outside parties can check future claims against.
A $2M Federal Award Will Build AI Tools to Catch Food-Safety Threats Before Outbreaks Happen
The University of Hawaii at Manoa is leading a $4 million NSF EPSCoR Track II award, in partnership with the University of Nebraska-Lincoln, to build AI-enabled tools that detect environmental threats to food-production systems before they escalate; UH's share is $2 million. The four-year project, which started September 1, 2026, will test the technology in aquaculture and beef-cattle production. It comes as public-health officials continue investigating foodborne cyclospora outbreaks that exposed gaps in detection speed. Why it matters: Food-safety failures are a recurring, low-visibility humanitarian burden, and this funds detection before the next outbreak rather than epidemiology after one — though it is a research award, not yet a deployed tool. The project's first field tests, in aquaculture and cattle production, are still ahead of it.
Upcoming Events & Opportunities
Active Disaster Monitoring (GDACS/OCHA)
- Nepal:** The Bhote Koshi glacial flood, active since 2026-08-26, has killed 1,344 people as of 2026-09-05, with roughly 275 Indian nationals still listed missing and NDRRMA reporting over 30,000 people affected; GDACS issued its highest-tier Red flood alert on 2026-09-01. Nepal is rebuilding the seismic-sensor, camera and satellite-link monitoring network at the China-border settlement of Timure after the original equipment was swept away; a separate investigation found the first disaster alert went out 38 minutes after the glacier collapse, with the Syabrubesi hydrological station offline mid-event.
- Alaska (United States):** A magnitude-6.3 earthquake struck 84 km SSW of Nikolski in the remote Aleutian Islands on 2026-09-03; no populated-area damage reports have surfaced. Included for magnitude threshold only, not for humanitarian response activity.
- Reuters, via Yahoo
- kathmandupost.com
- USGS earthquake feed (2026-09-03)
Sources: See individual stories above for full attribution.