AI for accountability

In Yemen, where years of conflict have left millions dependent on humanitarian assistance, getting cash to the right people at the right time can mean the difference between stability and real hardship. Each month, UNICEF helps 1.5 million households meet essential needs through its national cash transfer programme. But, as Gianluca Buono, Senior Coordinator, UNICEF Yemen, puts it, “when you’re working in a system of this magnitude, even minor irregularities — a missed payment, a duplicated record — can undermine trust and delay vital support for families.” 

Cash assistance is one of the most straightforward forms of aid, giving families the autonomy to decide how best to meet their needs. But behind that simplicity is an enormous amount of operational work. Payments move through multiple financial service providers, grievance calls come in daily from across the country, and verification often depends on data systems that don’t speak easily to one another. 

To bring about operational efficiency and mitigate risks, UNICEF Yemen is now using artificial intelligence (AI) to detect irregularities in its cash delivery process. Supported by the UNICEF Venture Fund, the work is one of the first attempts to apply AI-driven oversight to humanitarian cash assistance in a complex humanitarian setting. 

What makes this effort particularly notable is that the technology was developed in-house by the Country Office team itself, rather than by external contractors. “We didn’t want something that gets dropped in and forgotten,” said Gianluca. “We’re building capacity here, within our own team and within Yemen’s wider humanitarian context.” Staff in Amman and Yemen work together daily, training models, reviewing outputs, and figuring out how the tools can make real improvements to the programme. 

After an initial pilot finished in late 2024, the Yemen Service Centre began improving the model to detect anomoalies in the registration and enrolment of families into the cash programme. 

“AI doesn’t replace human judgement. It highlights what deserves a closer look, so our analysts can spend their time on the cases that matter most.” 

-Gianluca Buono, Senior Coordinator, UNICEF Yemen

For UNICEF, the Yemen experience reflects a broader shift in how frontier technologies are being used: not to replace people, but to support them in contexts where the stakes are high and the margin for error is slim. By learning from millions of data points, the models surface patterns that human teams alone could easily miss — not to remove the human element, but to reinforce it. 

In a country where trust is fragile and needs are immense, this work strengthens a simple but essential commitment: that assistance will be fair, accountable, and delivered with care.