How WFP and partners are using AI to support humanitarian response in emergencies: Three real-world examples
By Julia Dalibor
When disaster strikes, an earthquake levels homes, crops fail from drought or conflict forces families to flee, time is everything. The World Food Programme (WFP) must move quickly to assess needs, allocate resources efficiently and effectively and deliver life-saving assistance. But in many of the world’s most vulnerable regions, information is scarce, access is limited and conditions shift by the hour.
How can we make faster, more informed decisions when lives are on the line?
That question is at the heart of a partnership launched in late 2024 between WFP, the European Organization for Nuclear Research (CERN), the Luxembourg Institute of Science and Technology (LIST) and the Government of Luxembourg. The goal: to harness artificial intelligence (AI) in the service of humanitarian actions.
By combining WFP’s operational experience with scientific research and technical expertise from CERN and LIST, the partnership is working to develop responsible AI tools that can enable emergency responders to act faster and more effectively. While still under development, the first set of AI models is already being tested in real-world operational contexts and will soon be able to support early damage assessments, crop yield forecasting and strengthen data assurance.
Here are three promising examples:
1. Mapping earthquake damage within hours, not days
In the aftermath of a major earthquake, the first 72 hours are critical. But when communications are down and roads are blocked, responders often face challenges in assessing which areas have been hit hardest. Relying on satellite images to assess damages can be a solution. However, they can sometimes be obstructed by clouds or darkness, delaying critical decisions.
To address this challenge, WFP and LIST are co-developing an AI model that uses synthetic aperture radar (SAR) imagery, high-resolution images of the Earth’s surface, to detect structural damage, day or night, rain or shine. The model analyses before-and-after radar images to highlight where buildings may have collapsed and generates geospatial maps that could guide emergency teams more quickly and safely to the worst-affected areas.
Still in development and being tested on past earthquake data, the aim is to integrate the model into WFP’s SKAI platform to support real-time use in future emergencies. Work is also underway to expand its capabilities to assess flood and conflict-related damage, where fast, reliable information can mean the difference between life and death.
2. Predicting crop yields before the harvest
For smallholder farmers and entire communities across the Global South, one failed harvest can mean months of food insecurity. But anticipating these failures in advance, especially in a changing climate, is no easy task. Data often arrives too late to launch an early response.
Photo: © WFP/Arete/Kibuuka Mukisa
To get ahead of food insecurity, WFP and LIST are testing a machine learning model that estimates crop yields before the harvest. Using satellite imagery, soil data and weather patterns, the model is being trained to generate sub-national forecasts for key staple crops like maize and wheat.
The model classifies areas into categories like normal, of concern or critical, based on deviations from expected trends. These insights can be fed into WFP’s planning systems, such as the Humanitarian Data Cube, to guide early warning and anticipatory action.
The tool is still undergoing validation, but if successful, it could help governments and humanitarian actors prepare for climate-related shocks with increased efficiency and effectiveness, stretching limited resources to serve more people in need.
3. Strengthening trust in registration data
In emergency operations, registration data is the backbone of assistance. It includes basic information collected from people in need, such as names, household size, contact details and other key identifiers. This data helps humanitarian organizations verify who qualifies for support and ensure that food, cash or other aid reaches the right people.
But under pressure and without biometric systems, registration data can sometimes contain errors, such as duplicated entries, missing fields or unlikely combinations of information that may result in untended or duplicated assistance.
To help address this challenge, WFP and CERN are exploring the use of an AI model as a digital quality checker. From the outset, this work is being guided by strong data privacy, ethical safeguards and humanitarian values. The model, based on a type of neural network called an autoencoder, learns what “normal” registration data typically looks like and flags entries that appear unusual, such as households that are much larger than average, reused phone numbers or out-of-place demographic patterns.
The goal is not to replace human oversight but to support it, helping teams prioritize which records may need closer review. A user-friendly dashboard is also being explored to make the tool more accessible for field teams. While still in an early exploratory phase, this approach shows promise for strengthening data quality in complex, fast-moving emergency settings.
What’s next?
These examples are just the beginning. From early-stage pilots to field-ready platforms, the AI tools emerging from this partnership are grounded in one shared belief: technology alone cannot solve humanitarian challenges, but the right technology, co-created by people who understand both the science and the realities on the ground, can make a measurable difference.
What comes next? New AI models are already in development to anticipate extreme weather events, detect anomalies in digital cash transfers and Global Fleet data and monitor energy usage across humanitarian operations. The partnership is also exploring the use of federated learning, an approach that allows AI models to learn from decentralized data sources without compromising privacy. The goal remains the same: to equip those responding to emergencies with better, faster and more reliable tools, so they can reach people in need more efficiently and effectively when it matters most.
Learn more about the Strategic AI Partnership for Humanitarian Actions between the World Food Programme (WFP), CERN, the Luxembourg Institute of Science and Technology (LIST) and the Government of Luxembourg.
The WFP Innovation Accelerator sources, supports and scales high-potential solutions to end hunger worldwide. We provide WFP colleagues, entrepreneurs, start-ups, companies and non-governmental organizations with access to funding, mentorship, hands-on support and WFP’s global operations.
Find out more about us: http://innovation.wfp.org.
Subscribe to our newsletter.
Follow us on Twitter and LinkedIn and watch our videos on YouTube.
