Senior Researcher Mpho Moyo reflects on the Forum for Internet Freedom in Africa (FIFAfrica) session, AI Sovereignty for African Countries and Communities, co-hosted by Research ICT Africa, The Global Centre for AI Governance, Digital Action, Qhala and Masakhane African Languages Hub. The discussion focused on what sovereignty means for Africans and what it takes to achieve it legally, technically, and collectively.
Structural injustice across the AI value chain
The Just AI Framework of Inquiry was developed to guide both research and regulatory reform to ensure AI advances public interest outcomes across economy, ecology and society. It does this because, on a continent where power and opportunity are deeply unequal, and AI deployment only threatens further exclusion – particularly for those facing multiple intersectional inequalities, such as women and persons with disabilities – research and regulatory reform must look beyond aspirational ethics and responsibility, towards actionable mechanisms for embedding justice across the entire AI value chain.
Exploring AI impacts across the entire value chain, the Just AI Framework was developed in response to the dire need for African governance models that can account for diverse democratic systems, development priorities, and epistemic traditions. With this in mind, it identifies five interconnected governance challenges across Africa:
- Legal frameworks remain shaped by inherited colonial and global norms, with limited protections for those affected by AI’s reorganisation of markets;
- Data is frequently treated as a commodity, severing its connection to lived experience and claims for consent, compensation, and remedy. As an extension of our physical selves, more work must be done to give people agency over how their data is used, and the value it generates.
- Epistemic injustice arises when dominant languages and external approaches determine whose knowledge counts. Adding African data to a model does not give communities influence over its purpose or design. Extractive value flows compound this imbalance, where African data and labour generate value while ownership, profits and decision-making remain concentrated elsewhere.
- Environmental injustice follows when communities bear energy, resource and waste burdens without fair benefits or a say in the process.
- Technical performance alone cannot establish justice; we must judge progress by whether AI reduces barriers, improves livelihoods, protects workers, and enables redress. In this context, material outcomes matter.
This Just AI Framework became a central point in the conversation, offering an alternative to governance models borrowed from elsewhere in the world.
What does sovereignty mean for African people?
The session emphasised that AI sovereignty means having the power and capacity to decide how AI serves our societies. For nations, this means evaluating AI systems, setting conditions for their use, and enforcing local rights when harm occurs. For individuals, this means having a stake in the systems AI creates, a say in who can use your data and knowledge, and the ability to contest decisions you disagree with.
As it stands, however, most Africans lack agency over their data, and consequently, remain excluded from AI’s benefits. During the session, Masakhane’s Koliwe Majama presented research that supported this point, framing language inclusion as a matter of epistemic justice. She asked: whose language and knowledge does AI recognise, and whose does it overlook? When a tool cannot read or respond to local languages, it excludes people from information and public services, and prevents local innovators from exploring potential solutions to on-the-ground needs. When AI is deployed in Africa, Masakhane argued, communities should actively help identify its purpose, effectiveness, and application. Likewise, when local communities’ data is used, they should be recognised and fairly paid for the knowledge and labour they provided.
Having data within your country’s borders does not immediately grant you access to its value, or agency over its use. To gain AI sovereignty, we need more than data production. We need data stewards: skilled individuals, independent research organisations, and well-resourced institutions with the capacity to implement local, rights-respecting conditions. Likewise, we need laws with effective enforcement mechanisms and relevant channels to investigate complaints when things go wrong. Without them, sovereignty becomes an aspiration more than an action plan.
Cooperation and interdependence
African countries can strengthen this capacity together. Dr Wanjiru Mburu of Qhala proposed shared computing resources, pooled investment and collaborative AI training fellowships to help countries combine strengths and negotiate better terms with providers, while ensuring that smaller countries have a fair say and access to the resources they helped build.
Within this cooperative model, trusted data governance becomes essential. The African Union Data Policy Framework offers a starting point for rules adapted to local conditions. It argues that agreements must clarify how data is used and protected, and who responds when harm crosses borders. Likewise, it calls for democratic oversight and attention to the environmental costs borne by surrounding communities, aligning well with the same calls for environmental justice within the Just AI Framework of Inquiry.
African governments need to cooperate to finance the capacity to develop and govern AI using trusted data governance frameworks, and pool resources where shared investment meets common needs. Strategies require sustained funding for skilled staff, functioning institutions and oversight alongside infrastructure and maintenance. This financing must be built into implementation plans and public budgets. External support should reinforce priorities defined by African institutions and affected communities, with knowledge transfer and long-term capacity building that strengthen their ability to make and sustain their own choices.
Ultimately, this session underscored that people affected by AI should help shape decisions from the outset, including whether AI is the right response to their needs. Women, rural communities, persons with disabilities and other marginalised groups need meaningful opportunities to participate, supported by accessible information and relevant local language content. Institutions should explain how this input influenced their decisions and ensure people can challenge harmful outcomes, request a review, and receive an effective remedy.
Centring Justice in AI governance systems
RIA’s Just AI Framework of Inquiry provides a way to examine how power operates across the AI value chain and whether governance can redress the inequalities it sustains. By connecting human rights with economic and data justice, democratic oversight, care and sustainability, it asks whose knowledge shapes AI, who captures its value and who bears its costs. Applying this framework means looking beyond policy commitments to examine whether institutions can redistribute benefits, protect individual and community rights, and provide effective remedy. In this context, justice requires changing the conditions that produce exclusion, and addressing the harms of particular systems.
Because Africa’s AI ambitions must be measured according to just outcomes. Sovereignty gains substance when marginalised people can shape decisions, share fairly in the value their knowledge and labour create, and hold institutions accountable. To truly harness AI’s potential on the continent, cooperation must strengthen that power across borders, with safeguards against the concentration of resources and authority among already powerful actors.
After FIFAfrica, our task is to make these responsibilities binding through investment decisions, procurement and partnerships, and better resourcing for the institutions that enforce them. Likewise, we must commit to judging progress by AI’s ability to redress structural inequality and by whether those who bear its risks can demand change.
