Every meal counts: How WFP is using AI to reach more people, faster
The World Food Programme (WFP) is renowned as a frontline responder when conflict erupts or disasters strike, moving rapidly to deliver emergency assistance.
In emergencies, families can lose everything in days. WFP teams must act fast, often in crowded, fast-moving settings where people are displaced, records are incomplete and needs change by the hour. In these contexts, some people in need might not receive the necessary assistance.
“Moving fast and with accuracy isn’t a luxury – it’s essential.” – Magan Naidoo, Chief Data Officer
“When resources are limited, the way we manage information can determine how many people we are able to reach,” explains Magan Naidoo, WFP's Chief Data Officer. “Moving fast and with accuracy isn’t a luxury – it’s essential.”
That is why WFP works carefully to collect, store and manage information on who needs assistance – helping ensure people receive the right support and that resources are distributed fairly and accurately. For years, this has included manually checking fingerprint records and biographic data to identify possible duplicate records.
Now, WFP is among the first UN agencies to use photo-based technology and artificial intelligence to analyse large volumes of images in seconds, with our Enterprise Deduplication Service (EDS) helping identify duplicate records much faster than previously.
Why duplication can happen in emergencies
Duplicate records are a known risk across the humanitarian sector, particularly in fast-moving emergency settings.
This is because people may move frequently to escape violence or search for food, families may be registered under different spellings or languages, and responses often begin before systems across organizations are fully synchronized.
“Duplicate records can have real consequences: some people may receive more than intended, while others who need assistance risk being missed,” says Matthew Dee, Deputy Director, Supply Chain & Delivery.
“When needs are overwhelming and resources are limited, getting the details right can make the difference between whether someone receives the food they need – or not.”
How AI plays a key role in reducing duplication
EDS uses artificial intelligence to help identify possible duplicate records when people register for food assistance. It analyses information such as names, photos and – in the future – biometric data like fingerprints.
When two records appear to belong to the same person, the system flags them for review. A staff member then checks the information and makes a final decision, with human judgement remaining central.
"People get what they are entitled to – no more, no less." – Jacqueline Flentge, WFP’s ad interim Country Director in Mali.
“The tool doesn’t decide who gets assistance,” says Jacqueline Flentge, WFP’s ad interim Country Director in Mali. “Instead, it helps our teams see more clearly, so people get what they are entitled to – no more, no less.
”Beyond duplication, it also helps identify data-quality issues that staff then act on, strengthening the reliability of, and thereby trust in, WFP operations.”
“By reducing duplication and improving accuracy, EDS helps strengthen fairness for the people WFP serves.”Food is loaded at a warehouse in Kabul.
Results: fairer and fast assistance, measurable savings
The cost benefits of greater accuracy are already visible. In a pilot in Mali in 2025, EDS helped save more than US$431,000 in six months by reducing duplicated assistance. The solution is projected to save at least US$4.7 million in 2026 as it is scaled globally.
At an estimated cost of 70 cents per meal, that saving is equivalent to around 6.7 million additional meals.
“Every dollar saved through better data is potentially a dollar that can be turned into food,” says Flentge.
Moreover, the tool is saving staff time spent analysing data. Tasks that once took weeks – manually comparing spreadsheets and sorting lists – can now be completed in hours.
The system is also up to 50 percent cheaper than other biometric solutions, partly because it is built on open‑source AI models and does not require licence fees.
Responsible use of AI ensures data privacy
Data protection is extremely important to WFP. The system has been reviewed by WFP’s experts and through external audits, to ensure compliance with data protection and privacy policies and standards. Before being implemented in new contexts, it undergoes data privacy assessments and data protection analyses.
The system enables less intrusive and more culturally sensitive ways of collecting information, as Flentge explains: “We can now help ensure the right people receive the right assistance, through face recognition that works without people having to remove culturally important items such as veils or turbans.”
“This improves the experience of those we serve without slowing down assistance – supporting dignity, respect and efficiency.”
Crucially, while AI supports operations, decisions always rest with staff.
What’s next?
The tool has been deployed in Mali and piloted in Afghanistan, Burkina Faso, Cameroon, Mozambique, Niger, Somalia and Uganda.
When needs are rising and resources are limited, the best innovations are the ones that help humanitarian teams reach more people, more quickly, in a way that ensures their dignity.