Reinventing Humanitarian Procurement in the Age of AI
- Categories AI, Case Studies
- Date April 30, 2026
Reinventing Humanitarian Procurement in the Age of AI
How algorithmic systems are infiltrating humanitarian aid — and what to do about it
Only 8% organisations integrated AI
Billions in humanitarian procurement annually
Analysis and critique by EvalCommunity: This case study presents an independent analysis of the Access Now report “Reinventing Humanitarian Aid Procurement for the Age of AI” (March 2026) by Giulio Coppi. It reflects the perspectives of EvalCommunity on strengths, gaps, and recommendations for M&E and procurement practice. EvalCommunity does not claim ownership of the original content. Readers are encouraged to consult the original source for complete context and methodology.
Background: The AI Adoption Gap in Humanitarian Action
In recent years, humanitarian organizations have increasingly explored the use of AI and algorithmic systems to improve efficiency, decision-making, and service delivery. However, adoption has been markedly uneven. According to the Access Now report, based on surveys of over 2,500 humanitarian workers, 93% of staff have experimented with AI tools, yet only 8% of organizations have formally integrated AI systems. This gap highlights a critical issue: AI is spreading informally through “shadow AI” rather than through structured, governed processes.
At the same time, humanitarian organizations are becoming mediators of trust between vulnerable populations and private tech companies, introducing new risks related to data protection, ethics, and accountability. The report warns that AI is entering the humanitarian space “mostly by the backdoor, often without safeguards.”
The Core Challenge: Unstructured AI Integration
A large international NGO operating in crisis-affected regions faces a fundamental problem: How can the organization use AI tools to improve operations while ensuring protection of beneficiary data, alignment with humanitarian principles, and compliance with ethical and human rights standards?
Key Constraints:
- Limited internal AI expertise
- Heavy reliance on Big Tech platforms (cloud-based tools)
- Procurement processes focused on cost, not long-term risk
- Increasing use of AI tools by staff without oversight (shadow IT)
This leads to unstructured AI adoption, where tools are integrated through software updates, add-ons, external platforms, or individual staff accounts without formal evaluation or governance. The report identifies four pathways for AI integration: market-driven (purchasing/licensing), organisation-driven (internal development/partnerships), user-driven (shadow IT), and supplier-driven (major/minor updates).
AI Use Cases in Humanitarian Settings
AI chatbots for beneficiary questions, automated translation tools
AI-assisted reporting, proposal writing, knowledge management systems
NLP for survey analysis, predictive analytics for early warning systems
Medical supply forecasting, biometric verification, cash transfer programming
Risks Identified in the Report
AI tools are externally owned, constantly updated, and not fully auditable. Organisations lose visibility over system evolution.
Biased outputs, data misuse, harmful recommendations. Case study: chatbot upgrade introduced harmful outputs without user consent.
Humanitarian actors depend on few Big Tech providers who may shift priorities (e.g., defence contracts). Limited negotiation power.
Traditional procurement focuses on price and static products, but AI systems are dynamic and continuously evolving. Systems are not fit for purpose.
Solution: Strategic AI Governance & Procurement Reform
The report recommends shifting from transactional procurement to a strategic AI governance approach, moving from “buying tools” to “managing digital ecosystems.”
- Establish AI Governance Framework: Define approved vs. prohibited AI tools, introduce internal usage policies, create ethical guidelines for staff.
- Introduce Human-in-the-Loop Systems: AI outputs reviewed by human experts; no automated decision-making for high-risk areas.
- Strengthen Procurement Process: Include vendor risk assessments, human rights due diligence (hHRDD+), and continuous monitoring.
- Promote Open & Trusted Alternatives: Preference for open-source or auditable systems; use of sandbox environments for testing.
- Capacity Building: Training staff on AI risks and responsible use; reducing reliance on shadow AI.
Key Lessons for M&E and Procurement Professionals
If your organisation has no AI policy, you are likely already using AI informally.
AI risk starts at vendor selection and contracts, not at deployment.
Evaluate AI for data protection, fairness, and accountability — not just efficiency.
Most successful AI applications are narrow, task-specific, and low-risk.
AI should augment, not replace, human decision-making.
The “SAFE AI” Model for M&E and Procurement
Discussion Questions for EvalCommunity Users
- Is your organisation using AI formally or informally? How would you know?
- Do you have a process to evaluate AI tools before adoption — including updates to existing software?
- How do you ensure AI systems respect data protection and ethics throughout their lifecycle?
- What role should M&E teams play in AI governance, procurement, and vendor assessment?
- How can smaller NGOs and local organisations reduce dependency on Big Tech providers?
Conclusion
AI offers real opportunities for humanitarian and M&E work — but only if adoption is intentional, risks are managed, and governance is strong. Otherwise, organisations risk losing control over systems, exposing vulnerable populations, and reinforcing digital inequalities. The report calls for a shift from compliance-focused to strategic procurement, integrating zero trust principles and continuous environmental scanning into digital supply chain management.
Recommendations Summary
Establish forums for digital procurement reform, fund open-source alternatives, demand open procurement policies.
Rethink governance from transactional to strategic, update contractual terms, develop red lines for vendor vetting.
Replace opt-out with opt-in for AI features, restore human rights teams, mandate regular hHRDD+ assessments.
Advocate for local tech funding, pressure INGOs to support non-Big Tech alternatives, run customised audits.
References & Sources
- Coppi, G. (2026). Reinventing Humanitarian Aid Procurement for the Age of AI. Access Now. accessnow.org
- Access Now. (2024). Mapping humanitarian tech: exposing protection gaps in digital transformation programmes.
- NetHope. (2025). Humanitarian AI Code of Conduct.
- UN Working Group on Business and Human Rights. Corporate human rights due diligence guidance.
Disclaimer: Analysis and critique by EvalCommunity. This case study is an independent analysis for educational and policy guidance purposes. EvalCommunity does not claim ownership of the original Access Now content. Readers are encouraged to consult the primary source for complete methodology and context. The original report is published under CC BY 4.0 license by Access Now.
Published by EvalCommunity Academy. Based on Access Now research (March 2026) under CC BY 4.0. For professional M&E training on AI governance and procurement, visit EvalCommunity Academy.
The courses and articles have been developed by an experienced team of evaluators and software developers under the guidance of Fation Luli. The EvalCommunity Academy combines practical expertise in Monitoring & Evaluation with cutting-edge AI technologies to provide high-quality, accessible learning experiences for professionals around the world.