An AI Safety Report for the Global South, by the Global South
AI safety is a global conversation, but it has not yet become a truly global project. The frameworks being built to evaluate AI systems, the standards being set to govern them, and the institutions being created to oversee them have largely been designed in, and for, a small set of high-income countries. The Global South — its languages, institutions and communities — still remains at the margins of the larger AI safety discourse.That needs to change. And it needs to change with evidence.The Global South AI Safety Report is the Global South Network for Trustworthy AI’s contribution to that change. The report will serve as a living, rigorous reference point on AI safety in the Global South: examining how AI systems are being built, deployed, used, and maintained across different contexts, and identifying the safety concerns, governance gaps, and structural risks that emerge at each stage. Moving away from abstract risk taxonomies, the report seeks to illustrate how AI safety risks are actually emerging in practice across the Global South — through real-world deployment pathways, institutional arrangements, infrastructural dependencies, and everyday experiences of AI systems.Why This Report, Why NowGovernments across the Global South are building national AI strategies. Multilateral forums are negotiating the frameworks that will shape AI governance for decades. AI Safety Institutes are developing evaluation standards intended to be applied globally. All of this is happening with very little systematic evidence on how AI systems actually perform (and fail) across the Global South.The result is a compounding risk: safety frameworks that do not translate to the environments in which they are instituted and communities exposed to harms that aren't being measured or named.This report attempts to confront this risk head-on by providing a comprehensive evidence base on how AI harms materialise and are experienced in the Global South, and by proposing contextualised recommendations for addressing them.What the Report Will DoThe report has five core objectives: (i) to examine AI diffusion trends in the Global South; (ii) to identify and categorise AI safety harms in Global South contexts; (iii) to evaluate where existing safety frameworks fall short; (iv) to define the infrastructure needed to address these gaps; and (v) to translate findings into targeted advocacy with governments, technology companies, and global governance bodies.Research will unfold in two phases. An interim output will be ready ahead of the UN Global Dialogue on AI Governance in July. The full first edition will be published in time for the AI Summit in Geneva in 2027.Led By Expertise, Grounded in ContextThe report is co-chaired by Dr Kalika Bali (Microsoft Research India) and Dr Vukosi Marivate (University of Pretoria & UN Scientific Panel on AI) — two respected voices in AI research across the Global South. Their leadership ensures that the work is anchored in deep technical expertise and genuine regional understanding.Guiding the research is an expert advisory panel drawn from across the majority world:
Dr Andre Carlos Ponce de Leon Ferreira de Carvalho, University of São Paulo & Alan Turing Institute
Armando Guio-Espanol, Organisation for Economic Co-operation & Development & Berkman Klein Center
Dr Balaraman Ravindran, CeRAI, IIT-Madras & UN Scientific Panel on AI
Cecil Abungu, University of Cambridge & ILINA Program
Dr Chinasa T. Okolo, Technecultura; United Nations & the World Bank
Elina Noor, Carnegie Endowment for International Peace
Jamila Venturini, Derechos Digitales
Associate Professor Jonathan Shock, University of Cape Town AI Initiative
Dr Jerry John Kponyo, Kwame Nkrumah University of Science and Technology
Dr Jun-E Tan, Khazanah Research Institute
Maria Paz Canales, Global Partners Digital
Dr Monojit Choudhury, Mohamed bin Zayed University of Artificial Intelligence
Dr Payal Arora, Utrecht University & Inclusive AI Lab
Dr Rachel Adams, Global Center on AI Governance & Leverhulme Centre for the Future of Intelligence
Ridwan Oloyede, Tech Hive Advisory
Dr Rita Orji, Dalhousie University & UN Scientific Panel on AI
Sabhanaz Rashid Diya, Tech Global Institute
Dr Stephanie Kasaon, ActionLab & DAIvolve Technologies
Dr Sunayana Sitaram, Microsoft Research India
Dr Urvashi Aneja, Digital Futures LabWho This Is ForThe report is designed for policymakers navigating AI procurement and regulation, AI Safety Institutes seeking to expand their global evidence base, multilateral delegations shaping international governance, and technology companies operating across diverse markets.If AI safety is to mean anything in practice, it has to mean something everywhere. This report is how we begin to make that case: rigorously, consistently, and from the Global South outward.