In the debate on localisation, the claim that local players “lack capacity” actually points to an administrative skills deficit. Generative artificial intelligence can fill this gap, provided that its potential is not limited to satisfying funders, but serves as a source of empowerment.
Ten years after the World Humanitarian Summit and the striking of the Grand Bargain, under which signatories undertook to allocate at least 25% of aid funding directly to local and national players by 2020,[1]For further information, see, for example: The Inter-Agency Standing Committee, The Grand Bargain (Official website), https://interagencystandingcommittee.org/grand-bargain ; or, from the specific … Continue reading the outcome is indisputable: currently, these organisations only receive between 3% and 5% of overall funding. This ongoing gap, compared with international players, illustrates a collective failure to deliver on the promise of localisation.
The argument most often put forward to account for this situation is the “lack of capacity” of local organisations. As underlined in a Canadian study, “when problems of capacity arose, these were generally related to the ability of local partners to manage the compliance requirements of donors and funding.”[2]Julia Rao, Rapport des résultats de l’étude : engagement des organisations canadiennes de développement international envers la localisation, janvier 2023, … Continue reading However, these organisations clearly show they are able to design and implement effective programmes. In other words, the problem seems to be a deficiency less in operations than in administrative compliance.
This distinction is fundamental. What funders call a “lack of capacity” actually refers to a shortage of administrative infrastructure to ensure local players’ accountability, financial management and compliance with their policies. In this way, international organisations, who often have specialised teams for these functions, have a structural advantage.
It is precisely in this area that generative artificial intelligence (AI) offers an opportunity. By reducing the barriers associated with compliance and administrative processes, it could help, at least in part, to redress the imbalance in the conditions for accessing funding.
“Lack of capacity”: an institutional structure
Since 2016, many tools have been developed to evaluate the capacities of local players and identify improvement priorities: the CALP Network’s Organizational Capacity Assessment Tool (OCAT),[3]C. Mike Daniels, “Organizational Capacity Assessment Tool (OCAT) – User Guide”, CALP Network, January 2016, … Continue reading Tearfund’s Disaster Management Capacity Assessment Tool (DMCA),[4]Tearfund Learn, Disaster Management Capacity Assessment (DMCA) Tool, 2019, https://learn.tearfund.org/en/resources/tools-and-guides/disaster-management-capacity-assessment-tool and the Preparedness for Effective Response (PER) strategy of the International Federation of Red Cross and Red Crescent Societies.[5]IFRC, Preparedness for Effective Response Leaflet and Case Studies, 8 January 2023, https://www.ifrc.org/document/preparedness-effective-response-leaflet-and-case-studies
I have had the opportunity to lead several assessments using the latter tool, in particular in South Sudan. One observation kept coming up, which was that the gaps identified rarely related to the ability to intervene in the field, but rather to support functions: policies (for example, in child protection, supplies or combatting corruption), procedures, monitoring and evaluation, resource mobilisation, or financial management. Without specialised staff dedicated to fundraising, relations with funders or communication, these organisations find themselves in a vicious circle: by concentrating human resources in operational positions, they have fewer people available to meet the administrative requirements for obtaining funds. Conversely, my experience with international organisations tells a very different story: their expertise in this area (with specialised teams often backed by consultants) allows them to put together applications perfectly matched to the expectations of funders.
The imbalance is similar for reporting, especially given that the information must be collected and reformatted to meet various requirements, often duplicative (40% to 59% of content is found in several reports).[6]Erica Gaston, “Harmonizing Donor Reporting”, Global Public Policy Institute, February 2017, https://gppi.net/assets/Gaston__2017__Harmonizing_Donor_Reporting.pdf On average, the local organisations studied in the Global Public Policy Institute survey[7]Ibid. had to file forty-five to sixty-five annual reports to secure their funding, taking up a considerable amount of time that could not be spent on direct aid.[8]Ibid. A Norwegian Refugee Council internal audit showed that financial data being re-entered in different formats can represent up to 11,000 hours of extra work for the organisation. Resources are needed to meet funders’ requirements, but these requirements themselves determine access to resources.[9]Ibid. Artificial intelligence could significantly lighten this administrative workload[10]Office for the Coordination of Humanitarian Affairs, Briefing note on Artificial Intelligence and the Humanitarian Sector, 17 April 2024, … Continue reading and thereby help implement localisation in concrete terms.
AI as a tool for enhancing support functions
In terms of localisation, one of the most striking examples in recent years is that of the Emergency Response Rooms (ERR) in Sudan. In a context where many international organisations have left the country, these community networks coordinate humanitarian aid by means of WhatsApp groups. Since the start of the civil war in 2023, their volunteers have helped more than 11.5 million people by carrying out evacuations and providing access to water, healthcare and food. To facilitate access to international resources, AI has gradually become a key source of leverage.[11]Ka Man Parkinson, “How are humanitarians using AI in 2026? The case for governance and local leadership”, Humanitarian Leadership Academy, 14 April 2026, … Continue reading As a member of an ERR stresses: “Since the war started, we’ve relied on artificial intelligence to respond to donors’ requirements.” This use is revealing: AI is not used primarily to upgrade operations in the field but to meet the administrative requirements of the international humanitarian system.
“The most intensive and innovative uses have emerged straight from local practices, notably in Kenya, Sudan and Bangladesh.”
This trend is confirmed on a broader scale. A Humanitarian Leadership Academy (HLA) study[12]Ka Man Parkinson, Madigan Johnson and Lucy Hall, “Artificial intelligence in the humanitarian sector: mapping current practice and future potential”, Humanitarian Leadership Academy, 2026, … Continue reading showed that AI adoption has not spread in a North-to-South flow. The most intensive and innovative uses have emerged straight from local practices, notably in Kenya, Sudan and Bangladesh. More than 80% of responders to the HLA survey came from countries of operations, seemingly at the forefront of innovation in the use of AI.
Gülsüm Özkaya, whose research focuses on AI-generated images from the point of view of those affected by crises, offers a helpful new perspective. She considers the opposition between international and local organisations can become a false dichotomy when speaking about AI:
“The main divide right now is not about being global or local. It’s about being digitally fluent and AI-aware. A local organisation that masters the use of AI tools can access the opportunities and create impact as effectively as the global giants.”[13]Quoted in Ka Man Parkinson, “How are humanitarians using AI in 2026?…”, art. cit. See also Gülsüm Özkaya, “Should we use AI-generated imagery in humanitarian communications? Spotlight … Continue reading
On the ground, these uses are, above all, practical. During a discussion with Dyanne Marenco, president of the Costa Rican Red Cross, she explained to me how AI tools available free of charge help them improve their support functions, in logistics, administration and communication. The organisation has thereby been able to develop a more ambitious communications strategy, increase its visibility (to the point of even featuring in the magazine Newsweek), and make its interactions with funders more professional. They are in the process of formalising use of these tools and are in the development phase of their policy.
“AI was a key element of resilience in coping with the budget cuts initiated by the Trump administration at the beginning of 2025.”
The organisations interviewed for this article had all used AI over the course of the past year, almost exclusively to draft proposals and reports, draw up logical framework templates, develop indicators or create media content. For them, AI was a key element of resilience in coping with the budget cuts initiated by the Trump administration at the beginning of 2025. In addition, real-time translation integrated into video conferencing platforms allows more members to participate actively in discussions with international partners, in this way reducing language barriers.
Apart from large language models (LLM), some initiatives are exploring more advanced applications. Chatbots are used, for example, to facilitate complex consultation or negotiation processes between different local organisations speaking different languages. Each participant can share their point of view individually and confidentially before the tool produces an anonymised summary of everyone’s position. This type of approach has several positive effects: it reduces disparities in proficiency in English but also helps rebalance gender dynamics within groups. By guaranteeing a more equitable representation of voices, these tools limit the dominance of certain types of contributors, often male, when speaking in and steering discussions.
In the cases studied, AI is not used to replace human interaction but rather to reduce time spent on paperwork. By reducing the cost of administrative functions, it allows local organisations to focus on what cannot be delegated to AI: interaction with communities.
In a system where access to funding largely depends on the ability to meet complex compliance requirements, this development could help reduce a major structural imbalance.
Limitations and risks
The limitations and risks associated with AI should not be overlooked, especially since its use is still largely informal, led by individuals rather than the organisations to which they belong. Only 23% of organisations actually have dedicated policies or training.[14]Ka Man Parkinson, Madigan Johnson and Lucy Hall, “Artificial intelligence in the humanitarian sector…” art. cit.
This lack of supervision poses a significant risk. According to a research centre at Stanford University, the number of incidents[15]An incident refers to “proven cases where AI systems have caused or nearly caused harm”, see “Artificial Intelligence Index Report”, Human-Centered Artificial Intelligence, 2026, p. 132, … Continue reading in AI use reported in 2025 rose by approximately 55% compared with 2024.[16]Yolanda Gil and Raymond Perrault (ed.), Artificial Intelligence Index Report, Standford University Institute for Human-Centered Artificial Intelligence, 2026, … Continue reading Timi Olagunju, involved in policies and governance related to AI, highlights a lack of governance as a major issue for the sector:
“The fact that AI policy is at a slow pace compared to the growth in use within the humanitarian sector is concerning. Governance frameworks provide the context in which AI can truly serve the public good.”[17]Rebecca Chandiru et al., “Beyond the hype: Ground truth on AI across the humanitarian sector”, Humanitarian Leadership Academy, 26 February 2026, … Continue reading
The implementation of policies, training or even specialised positions dedicated to responsible use of AI will be easier for organisations that already have considerable financial resources for their support functions. This could lead to a new imbalance: not only in access to tools that are more suitable than those of widespread commercial models, but also in their ability to manage the risks associated with their use. Several organisations[18]Costa Rican Red Cross, ICRC, UNHCR, Save The Children, NRC, ERR Soudan, Kictanet Kenya, African Institute for AI Governance and Ethics (AIGE), Maskhane, Digital Umuganda Rwanda, Data friendly Space … Continue reading are already working hard on developing training courses, risk management tools and guidelines for responsible use of AI.
It is nonetheless important to put into perspective the risks associated with the scenarios discussed in this article. Concentrating AI use first on administrative functions will reduce the most significant risks: sharing incorrect information with or about affected communities,[19]Ana Beduschi, “Harnessing the potential of artificial intelligence for humanitarian action: Opportunities and risks”, International review of the Red Cross, no. 919, June 2022, … Continue reading misidentifying affected areas,[20]Unitar, Fusing AI into satellite image analysis to inform rapid response to floods, 13 April 2021, … Continue reading confidentiality of personal data, poor selection of beneficiaries[21]Andrew Deck, “AI translation is jeopardizing Afghan asylum claims”, op. cit. and “hallucinations”[22]In the field of AI, we call a hallucination a false or misleading response presented as a certain fact, for example a bibliographic reference [editor’s note]. that can have real-life consequences for vulnerable individuals.
Conditions needed for a form of AI that benefits localisation
The true extent of the impact of AI on localisation may lie elsewhere. Even if AI ultimately allows local organisations to improve their compliance, reporting and administrative functions, there is no guarantee that this will lead to a significant increase in their funding from international funders. It is possible that the “lack of capacity” argument is less a real cause of the underfunding of local organisations than a convenient justification. If this is the case, the real obstacle is not technical but political.
Let us consider the worst-case scenario in which, regardless of progress made, traditional funders do not commit to localisation strongly. Indeed, the past ten years do not point to a more optimistic scenario. Yet this hypothesis seems all the more plausible against today’s backdrop of the multilateral agenda withdrawing and of international aid declining. In this context, a question comes to mind: are there any alternative routes for local organisations? And, if there are, can AI be part of them?
In many contexts, ways to raise funds already exist, through diasporas, community contributions or religious solidarity mechanisms, or even via mutual funds such as the NEAR foundation[23]See: https://near.ngo or the Resilio Fund.[24]See: https://resiliofund.org AI can play a pivotal role here.
Local organisations already use AI to develop their brand image and visual identity, and, increasingly, to create content and increase their visibility at a local and international level. But several other opportunities are currently emerging. Content generation tools, chatbots specialised in communication, or even donation management solutions provide an opportunity to devise more ambitious fundraising strategies, without needing large-scale investment in philanthropy departments. They pave the way for targeted digital campaigns, the development of local donation platforms, the use of crowdfunding and greater engagement of diaspora communities.
“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.”
This prospect is all the more relevant that more than 60% of humanitarian operations now take place in middle-income countries.[25]The author’s personal estimate based on data collected, for example: World Food Programme, WFP’s engagement in middle-income countries (2019–2024): Evaluation synthesis, 21 May 2025, p. 2, … Continue reading Yet it is also in these settings that we see a growth in the middle class, a rise in the numbers of the super-rich and an increase in donations and volunteering. In other words, the conditions for the emergence of local philanthropic ecosystems are already in place.
Though AI is largely employed today to improve compliance functions, its potential remains under-exploited in the development of alternative funding models. 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. On the other hand, AI has transformative potential if this tool can free local organisations from funders by making them more independent. Only on this condition will artificial intelligence truly become a driver for localisation of the humanitarian system.
Picture credit : Adrienne Surprenant / Collectif Item
Translated from the French by Fay Guerry
