5 Ethical AI Principles Every NGO Should Follow in Humanitarian Projects
As artificial intelligence (AI) transforms humanitarian action—from predicting refugee displacement to automating aid delivery—NGOs face urgent ethical challenges. Over 80 million displaced people rely on aid systems increasingly shaped by opaque algorithms, raising risks of bias, privacy violations, and exclusion :cite[6]. To navigate this landscape, NGOs must adopt ethical frameworks that align with global standards like UNESCO’s Global Digital Compact and UNHCR’s partnership with Arm on human rights-based AI :cite[1]:cite[4]. Below are five principles to guide responsible innovation.
1. Human-Centered Design
- Principle: Prioritize human rights and dignity in AI systems, ensuring they augment—not replace—human agency.
- Action Steps:
- Anchor AI tools in international human rights law, as advocated by the Cambridge study on humanitarian AI ethics :cite[1].
- Design systems that uphold humanitarian principles: humanity, neutrality, impartiality, and independence :cite[2].
“Life and death decisions should not be delegated to machines.” — UN Secretary-General’s Roadmap for Digital Cooperation :cite[2]
2. Refugee Data Sovereignty
- Principle: Ensure refugees retain ownership and control over their data.
- Action Steps:
- Store data locally (e.g., on servers at institutions like Gulu University, ASAT Labs) to comply with GDPR and avoid exploitation :cite[1].
- Avoid biometric systems unless strictly necessary, as highlighted by critiques of facial recognition in camps :cite[1]:cite[5].
3. Algorithmic Transparency and Explicability
- Principle: Make AI decision-making processes understandable to affected communities.
- Action Steps:
- Reject “black box” models. Use explainable AI tools [ASAT Labs Humanitarian Ai] to demystify outcomes, as urged in the Journal of International Humanitarian Action :cite[2].
- Publish audit reports, as demonstrated by New York City’s algorithmic transparency directory :cite[2].
4. Inclusive Co-Design with Affected Communities
- Principle: Involve refugees in AI system design to prevent algorithmic colonialism :cite[5].
- Action Steps:
- Conduct participatory workshops, like Uganda’s TeamUp program, blending community feedback with technical development :cite[1].
- Amplify marginalized voices, addressing the gender gap in AI ethics authorship (only 31% female contributors) :cite[5].
5. Rigorous Impact Assessment and Accountability
- Principle: Continuously evaluate AI systems for unintended harm.
- Action Steps:
- Adopt the AI Suitability Framework by NetHope, which includes ethical risk scans and mitigation strategies :cite[6].
- Establish third-party review boards, as proposed in OCHA’s predictive analytics catalog :cite[2].
Conclusion: A Call for Ethical Leadership
The humanitarian sector cannot afford to outsource ethics to tech companies. By adopting these principles—grounded in UNESCO’s toolkit and UNHCR’s partnerships—NGOs can pioneer AI that respects refugee rights while scaling impact. As NetHope’s Leila Toplic notes, “AI must be responsible, sustainable, and impactful” :cite[6]. The path forward demands humility, collaboration, and unwavering commitment to the communities we serve.
Further Reading
- UNESCO’s Global Digital Compact :cite[4]
- NetHope’s AI in the Humanitarian Sector Report :cite[6]