The global narrative surrounding AI remains trapped in a narrow, self-referential loop. Walk into any major international summit, and you will hear AI governance discussed almost exclusively through the lens of domestic innovation, market efficiency, and the enforcement of abstract, principles-based frameworks designed to mitigate immediate technological risk.
But at the 2026 Africa Just AI conference, a powerful, collective counter-narrative emerged. The consensus was clear, uncompromising, and urgent: dominant global frameworks fundamentally ignore the deeper structural inequalities inherent in the global AI value chain.
In an era of extractive AI, the underlying resource burdens – ranging from data harvesting to exploitative content moderation labour, and devastating ecological costs – are disproportionately outsourced to the Global South. Even more concerning is the growing dependency on monopolistic cloud infrastructures.
To transition to a truly inclusive global governance architecture, the international community must confront the core contradictions of this current matrix. We must look beyond localised risks and actively dismantle the structural asymmetries of the AI pipeline. With a renewed oversight model, mechanisms must be established to ensure that advanced AI capabilities are no longer subject to arbitrary, unilateral restrictions.
For Africa, this means moving beyond the passive adoption of foreign models. The age of extractive AI must end, the era of Just AI must begin.
Consultation to binding accountability
For too long, African AI governance has relied on passive advisory committees and symbolic stakeholder consultations. We must transition from mere consultation to binding accountability.
When algorithms exercise public functions – such as welfare allocation, policing, or credit access – they are not neutral technical tools; they are performing actions that fundamentally alter human lives. In Africa, we must formally classify these deployments as “administrative action.” By subjecting these systems to existing, rigorous legal standards of legality, reasonableness, and procedural fairness, we embed democratic values directly into the algorithmic architecture.
African scholars and practitioners are advocating for two concrete mechanisms to enforce this:
Citizen-led algorithmic oversight councils (CAOCs): Statutory bodies endowed with the explicit authority to suspend harmful systems and mandate technological redesigns.
Community-driven AI impact assessments (AIIAs): Moving past superficial technical checklists, these mandatory assessments incorporate local knowledge, ensuring systems reflect African epistemologies and philosophies such as Ubuntu, which prioritises collective well-being and relationality over the solitary, extractive efficiency of foreign models.
Regional sovereignty as a strategy
Digital sovereignty cannot remain an aspirational catchphrase; it requires operationalised regional roadmaps. We must leverage instruments such as the Malabo Convention to establish a continental baseline for privacy and robust cybersecurity enforcement. We need harmonised labour protections for the rapidly growing class of African algorithmic workers. Data sovereignty must shift from a defensive posture to an aggressive economic development strategy.
Instead of tethering our digital future to external compute monopolies, Africa must invest in sovereign infrastructure. This means building regional GPU-as-a-Service hubs and communal data trusts that treat data as a strategic collective resource rather than an extractive private asset. Local compute allows us to train models that actually perform for African languages, diverse faces, and unique economic realities, ensuring that the economic value generated by AI accrues directly to the communities providing the raw material.
Reclaiming our epistemic agency
At the heart of the current AI value chain lies a condition of “epistemic dependency.” African languages and knowledge systems are routinely treated by Silicon Valley as low-resource, invisible, or mere training fodder for Global North dominance. A justice-centred approach requires us to reject the passive goal of merely seeking “access” to unequal platforms. We must replace it with the active goals of agency, design, and control.
Language communities must maintain ownership and stewardship of their own datasets. This requires a definitive end to “openwashing,” where major tech firms use open-source licenses to extract uncompensated local labour. Legal frameworks must recognise and protect rights to community-generated data, guaranteeing equitable benefit-sharing and collective attribution.
Furthermore, true justice requires deliberate, structural policies that support women’s progression in AI through all stages – from early education to technical leadership and high-level policymaking.
The era of Just AI
The utility of the Research ICT Africa Just AI Framework extends far beyond discourse.
On the global stage, the Framework is the normative bedrock for Research ICT Africa’s engagement with the United Nations (UN) Global Digital Compact (GDC) and the upcoming UN Global AI Governance Dialogues. The Framework can be used to bridge the technical data governance work of the UN Commission on Science and Technology for Development (CSTD) with the broader AI agenda, ensuring data is treated as a fundamental instrument for accountability and human rights, not just a commodity. Amidst a global discourse dominated by the rapid rise of highly capable open-source models and the evolution of technical AI safety standards, Just AI requires that open-source AI is not used as a vehicle for digital colonialism, and that AI safety is not defined by how secure a system is for a tech conglomerate, but by how fair and self-determined it is for those impacted. The legal authority for communities to suspend harmful systems and mandate technological redesigns is urgent and vital.
For regional policymakers, the Framework provides the operational “how-to” that regional AI Strategies urgently require. Through convenings and Schools on AI governance, the Just AI Framework of Inquiry fosters engagement between national policymakers and African scholars, shaping Africa’s leading thinkers as they strategically build sovereign infrastructure and develop African agency and expertise. AI in Africa must live within a symbiotic digital infrastructure where the value generated by African data ultimately serves African people.
For civic tech, Just AI represents a shift from passive consumption to active control, an opportunity to disrupt power asymmetries, foster local open AI innovation and utilise local data to ensure the socio-economic opportunities of AI are captured locally rather than exported. The expanded Framework is a uniquely positioned governance model for local-language tech, defining pathways for building and safeguarding localised datasets and ensuring that sovereign, local-language AI models trained natively mitigate reliance on Global North infrastructure.
As an AI governance paradigm, Just AI is being used as a new curriculum for AI governance, in regional and global schools and leadership gatherings, in Africa and globally.
We are at a defining moment for African AI futures. We cannot measure success by the velocity of our technological adoption or the sheer efficiency of automated systems. Instead, we need to ask whether our AI strategies and governance enable transformative AI value in the public interest. We begin with the commitment to uphold economic and administrative justice, strengthen our democratic institutions, and ensure that our technological future is self-determined. Africa does not need to choose between participating in the global AI economy and protecting our sovereignty. We do need to choose, however, to respect our histories, empower our communities, and serve our collective futures.
