Artificial Intelligence

This project is now closed.

As Artificial Intelligence (AI) dramatically reshapes the world, the humanitarian sector must navigate a path that leverages AI's potential to enhance humanitarian practice while addressing its associated challenges and risks.

With global humanitarian needs at an all-time high - driven by prolonged conflicts, climate change, and increasing fragility and resources stretched thinner than ever, innovative solutions are urgently required to address shared challenges effectively.

What was the innovation?

Our collaboration between NetHope and the UK Humanitarian Innovation Hub (UKHIH) represents a structured and system-oriented effort to advance the responsible, scalable use of artificial intelligence across the humanitarian sector through the development of an AI Lighthouse. This initiative is designed not as a standalone tool, but as an enabling architecture that connects actors, consolidates learning, and accelerates the transition from fragmented experimentation to coordinated, evidence-driven adoption of AI.

How was this achieved?

Phase 1: Mapping AI-enabled projects for humanitarian impact

Emerging technologies like AI have the potential to revolutionise humanitarian action by expanding reach, improving efficiency, and amplifying impact. Despite the rapid increase in AI pilot projects, adoption across the sector remains fragmented. Limited publicly available information about these initiatives hinders collaboration, lesson-sharing, and the development of safe and reliable AI solutions.

To bridge this gap, we launched a project to map AI-enabled tools and projects supporting the humanitarian sector. This initiative aimed to equip humanitarian practitioners with the tools to:

• Procure safe and dependable AI models.
• Foster collaboration and build on existing solutions.
• Address global humanitarian challenges through innovative approaches.

We invited colleagues from both the humanitarian and technology sectors to contribute their insights to this growing open-access directory. By working together, we have aimed to foster future-ready innovations and partnerships to tackle the increasing scale of global humanitarian challenges.

Phase 2: Development of the AI Lighthouse

Workstream 1, implemented in partnership with the Technical University of Munich, focused on AI scoping and landscaping at a system-wide level. This included a comprehensive analysis of the humanitarian AI landscape, identifying key actors, use cases, and gaps, alongside structured stakeholder consultations to ensure alignment with sector priorities and operational realities. This process also closely aligned with other sector wide AI initiatives, ensuring coherence with emerging norms around safe, ethical, and accountable AI deployment.

Workstream 2 translated this strategic foundation into practice through catalytic grant funding aimed at learning from the scaling of high-potential AI applications. This included funding support to the International Rescue Committee for the development of Signpost AI, an open-source orchestration platform designed to deploy safe and contextually appropriate AI agents in humanitarian settings. In parallel, UKHIH supported the Norwegian Refugee Council in advancing CLEAR, an initiative that leverages AI for crisis learning, early warning, and anticipatory response. Across both projects, research was conducted to better understand the unique challenges, solutions, and strategies that characterise the scaling of AI tools within the sector. Together, these investments demonstrate how targeted and strategic funding can move AI solutions beyond pilot phases toward operational integration.

Workstream 3 strengthened our evidence base through a portfolio of case study grants. This workstream ensures that insights from real-world AI implementations were rigorously captured, comparable, and transferable across contexts. The result is a curated body of evidence that moves beyond anecdotal success stories to provide actionable learning on what works, under what conditions, and why.

Finally, Workstream 4 addressed a critical knowledge gap through the development of a Humanitarian AI Directory. Based on a sector-wide survey of AI initiatives, this directory provides a structured and regularly updated mapping of “who is doing what” in humanitarian AI. By increasing visibility, reducing duplication, and enabling strategic collaboration, it serves as a foundational public good for the sector.

Taken together, our partnership with NetHope through the AI Lighthouse initiative reflects a deliberate shift toward system-building: combining landscape intelligence, catalytic financing, applied evidence generation, and shared infrastructure to support a more coherent, accountable, and impactful humanitarian AI ecosystem.