Abdullah X & Fiza Siddiqui (Ashoka University) have posted Beyond Universal Metrics: A Global South-Centered Framework for Ethical AI Development on SSRN. Here is the abstract:
The rapid deployment of artificial intelligence systems has produced documented harms disproportionately affecting marginalized communities in the Global South, from algorithmic caste discrimination to exploitative labor practices in AI supply chains. This paper presents a comprehensive ethical framework developed through systematic analysis of empirical research on AI-related harms and existing governance approaches. We introduce ten foundational principles that address critical gaps in current AI governance: contextual fairness over universal metrics, transparent accountability with innovation protection, labor dignity in AI supply chains, data sovereignty with innovation access, environmental justice, epistemic justice, democratic governance, proportional regulation, security without surveillance, and information integrity. Our framework distinctively synthesizes technical AI safety research with social justice scholarship, centering perspectives from affected Global South communities while enabling responsible innovation. Unlike existing approaches that often impose Western epistemologies and governance models, we demonstrate how context-aware, community-centered principles can simultaneously advance AI capabilities and protect vulnerable populations. This work contributes to AI ethics literature by providing actionable guidance for organizations navigating rapid AI adoption while addressing systemic inequities embedded in current development practices.