AI could be a lever for localisation – but genuine partnership and investment is needed

In those stressful moments, using AI didn’t just help us work faster, it gave us confidence, relief and even a sense of hope.

Abdullah Azimi, a humanitarian practitioner in Afghanistan

In May 2025, we started a global survey to ask how humanitarians are using AI – and the results were not what we expected.

The Humanitarian Leadership Academy/Data Friendly Space research captured a snapshot of a sector caught between the turbulence of the Humanitarian Reset and the arrival of generative AI. This picture persisted in our 2026 follow-up survey.

This presents a paradox: individual uptake of AI tools is outpacing organisational readiness, including AI adoption and governance.

This is not a Global North-to-South AI adoption pattern. Usage in low-resource, low-connectivity settings, and among local and national actors, often exceeds levels in international, Global North organisations. The highest daily usage in our January 2026 survey occurred in Kenya, Sudan and Bangladesh.

This is a double-edged sword. Rapid, individually-driven adoption brings real data and governance risks but it also presents an opportunity for change.

I believe this moment offers a narrow, time-bound window to prepare for the AI era – and to reimagine systems in alignment with localisation processes.

Alongside the urgent need for sector-wide AI governance and literacy support, three themes have emerged from this sectoral listening exercise which require our collective focus now.

1. A question of framing: situate AI in the localisation agenda

AI should not be seen as separate from humanitarian principles. It should be inclusive, anticipatory and accountable.

Programme manager Shudarshan Hamal, Nepal

AI is often framed as an efficiency and innovation question. In the humanitarian context, it must be seen more holistically: as a localisation question; one which is owned by policymakers, NGO leaders and decision-makers, not just data and technology teams. Accountability to affected populations and do no harm principles must sit at its core.

This year we’ve seen developments within the sector with exactly this framing; namely, through coordination and standard setting. This includes the FCDO-funded SAFE AI governance framework, and the Inter-Agency Standing Committee’s development of its Framework for the Responsible Use of Artificial Intelligence in Humanitarian Action as part of the Humanitarian Reset.

Yet, within organisations our research shows overall stagnation on AI readiness measures, such as governance, adoption and training. This is unsurprising, given the funding shortfalls and compounding crises organisations are managing.

As the sector faces enforced structural change – regardless of AI – new operating models, adaptation and consolidation will happen. AI is inside that transition, whether the sector plans for it or not, given how common individual usage is.

NGO leaders redesigning systems and processes have a choice about whose priorities shape them. Co-creating with local and national actors offers an opportunity to re-think legacy systems and avoid top-down solutions. As Nour Arab, a humanitarian practitioner from Lebanon, observes:

We had this mistake with digital transformation – we spent millions of dollars trying to create digital products that nobody used…This is a golden time for us to realise that the power is no longer on the upper end.

The organisational decisions of 2026-2027 cannot exacerbate existing systemic divides. Ulrich Assouah, Managing Director of IFP Humanitarian Studies in Douala, Cameroon, argues:

We need to avoid a new digital inequality where only large international organisations benefit from AI transformation, while local actors remain behind due to lack of access, fractional training or funding.

2. Co-create equitable mechanisms and systems

A local organisation that masters AI tools can create impact as effectively as the global giants.

Gülsüm Özkaya, a young humanitarian leader from Turkiye

AI adoption does not necessarily equate to increased capacity. Currently, the use of AI across the sector often goes toward the heaviest administrative burdens; chiefly, resource mobilisation and donor reporting. Such burdens are felt most acutely by local and national actors, even during severe crises, including in Sudan.

This represents another growing challenge in the AI era to be actively mitigated. In an article for Humanitarian Alternatives, building on our research, humanitarian consultant Jean-Baptiste Lacombe Lavigne warns:

The impact of AI will be low if it is only used to improve compliance with the requirements of funders who, we can fear, will not fund localisation any more than they have done up to now.

The potential benefits of AI for local actors could be better realised by first relieving systemic pressures. This means processes which reduce duplicated effort, such as pooled funds, streamlined reporting and grant management systems, and due diligence passporting.

These mechanisms must be designed to truly enable local actors to free up capacity so they can leverage the benefits of AI, rather than simply raising the bar for other forms of compliance.

Dedicated and inclusive space and time is needed for design thinking, and to bring pre-AI and localisation processes into the reform era. Pace is critical, given the significant time lag between organisational policy action and the speed of AI developments. By way of illustration, between our 2025 and 2026 surveys, AI policy adoption barely shifted (from 22% to 23%).

3. Fund the infrastructure for shared learning and action

We need an artificial intelligence that speaks the language of the donor and the language of the village where I come from.

Ivan Toga from Uganda, who works with refugees

Technology cannot be imposed on contexts in the name of partnership. Participatory design, deployment and governance led by local and national actors is needed.

This requires properly funded and supported mechanisms to convene and support the operational levels of this humanitarian ecosystem through dialogue, shared learning, design and deployment of AI tools and systems.

These mechanisms must not be dominated by Global North actors, but designed with and for Global South actors – alongside technologists, funders and more – as a balanced capacity-sharing model.

The HLA and NetHope webinar I convened in July this year brought technologists and humanitarian practitioners into the (virtual) room to work with a shared definition of localised AI. This was presented and understood not as a tech stack question, but as an AI system designed with, and ultimately owned and governed by, the communities it serves.

That shared understanding lays the foundations to build from, together. This is a step that practitioners and panellists alike reflected can be missed in reform and localisation processes.

As the sector reforms under pressure, AI shouldn’t just help it survive as it is. This is the moment to build the system we actually want.

This is not advocacy for increased AI adoption or technosolutionism. It’s a call for the purposeful partnership and intentional investment needed to build responsive, equitable systems, with localisation baked into the emerging new blueprints.