Day 1 of the Just AI Conference was, by design, an academic grounding. Day 2 asked a harder question: now that we know what Just AI requires, what does it actually take to build it?
As Dr Kola Ijasan (Research Director, Research ICT Africa) reflected in his opening remarks, that we are no longer debating whether AI should be ethical, responsible or just. We know the answer. What matters now is something more demanding: embedding justice into our regulatory responses, and determining who bears responsibility for enforcing them. When injustice runs across an entire value chain — infrastructure, data, skills and capital — dismantling it requires close coordination, multistakeholder reinforcement, clear mandates, and participatory political will. These are the things that stuck with us.
To govern AI justly, we cannot just be consumers
Africa currently hosts less than 2% of global data centre capacity, yet it accounts for 20% of the global population. When Pamla Gopaul (Programme Head, Africa Policy Bridge Tank, Economic Analysis and Foresight Unit, AUDA-NEPAD) said this, she reminded us that Just AI is not only about how we govern new technologies, but also who governs the systems that make them possible in the first place. Dr Jacob Munodawafa (CEO, Southern Africa Telecommunications Association) linked this lack to concrete harms such as misidentification by AI-assisted facial recognition systems and failures in AI-assisted medical diagnoses. These case studies demonstrate that without African inputs, effective AI usage in essential sectors becomes impossible. Dr Bhekani Mbuli (Researcher, University of Johannesburg) took things a step further, asserting that Africa’s core problem is not a lack of ideas, policies, or strategies. Rather, it is poor execution. The Continental AI Strategy, the African Continental Free Trade Area, and the African Union Data Policy Framework are proof that the continent already has an existing and layered set of regulatory instruments to guide AI development and deployment in the continent, for the continent. What is missing, however, is grounded, coordinated action to bring them to fruition in diverse regions.
AI stakeholders must move beyond declaration to enforcement
Session 2 brought together regulators and AI policymakers from the SADC region, who offered a candid account of what AI policy processes look like from the inside. Dumisani Sondlo (AI Policy and Governance Lead, Department of Communications and Digital Technologies) began by sharing his encounters navigating not just policy logic but also political dynamics, personalities, and institutional inertia. In doing so, he evidenced that the success of AI strategies and agendas is highly dependent on institutional readiness and capacity. Dr Mukelani Dimba (South African Information Regulator) echoed these thoughts, describing the difficulties of navigating the recent Generative AI scandal on Grok, which called for the mobilisation and application of multiple different policy instruments, including the Protection of Personal Information Act, and cybersecurity laws. In such dynamic environments, where data justice, privacy, and cybersecurity are threatened by AI developments, coordinating across departments and tools becomes complex. Chipego Hakalima (Deputy Data Protection Commissioner, Registrations and Licensing, Data Protection Commission, Zambia) emphasised the importance of clarifying which instruments and actors can best be utilised in such crises, especially amid political change. For example, she questioned how Zambia’s Data Protection Commission might operate when Zambia’s national AI policy expires in 2026. Without any concrete, enabling legislation, and a soon to be desolved Cabinet in the upcoming elections, government actors are challenged to respond to rights-violating events with limited resources.
Justice cannot be borrowed – it must be built
Chenai Chair (Director, Masakhane African Languages Hub) opened the last panel by arguing that the language of justice has itself been co-opted, with the vocabulary of openness and inclusion now appearing in corporate AI frameworks and PR stunts, thanks to its normalisation by multistakeholder communities and advocates. Her points echoed those made by Joanne D’Cunha (Programme Manager, Centre for Communication Governance, National Law University Delhi) in an earlier session; reflecting on the India AI Impact Summit, D’Cunha argued that while the summit and its accompanying global AI governance processes are making genuine progress in terms of expanding participation and recentring marginalised development priorities, the optics of inclusion and multistakeholderism are often performative, failing to elicit authentic coordination and co-development with dominant players. Dr Annette Hubschlë (Senior Researcher, African Hub on AI Safety, Peace & Security, University of Cape Town) argued the same, especially from a scholar’s perspective, noting how few social science and humanities players are invited into these rooms to incorporate a language of care and community into industry-wide decisions. Dr Alison Gillwald (Emeritus Distinguished Fellow, Research ICT Africa) regarded these observations from a political economy perspective, saying that Africa cannot win by funding and mobilising its tech advancements through and with the very institutions that those strategies are meant to hold accountable. She made the case that genuine multilateral solidarity is not one option among many; it’s the only one. A contribution from the floor extended the argument to Geneva, where the terms governing knowledge and intellectual property are still being written – and where Africa must show up not as respondent, but as co-author.
The execution gap must take priority in the future of the Just AI agenda
Africa does not have a shortage of vision for just AI governance. It has a shortage of mechanisms to translate that vision into action. Strategies without enabling legislation, resources, or partnered support cannot be enforced. What we need now is coordinated political will. For Just AI’s future to become a reality, we need to build the institutions, coalitions, and enforcement mechanisms it requires to benefit all equitably and with care.
