Framed as a pivotal moment for AI for All, the United Nations Global Dialogue on AI Governance (UNGDAI) in July 2026 successfully captured the attention of multistakeholder communities, following months of cross-geographic engagement. However, this inaugural session also served as a stark reminder of the ongoing contest for AI futures. Currently, the discourse remains trapped in artificial silos: high-level policy declarations adrift in abstract, insulated processes, while technocratic solutions appear unencumbered by the urgent need to balance human rights with economic imperatives.
Continuing on this trajectory risks further entrenching a neo-colonial value chain where the data and labour of Global South populations are extracted or harvested. What we need as a global collective is a (radically) re-balanced governance Agenda, for just and equitable AI beneficiation frameworks for All.
A crucial framing input for the UNGDAI was the UN’s Independent International Scientific Panel on AI Preliminary Report, released on 1 July 2026. Co-chaired by Yoshua Bengio and Maria Ressa, the 40-member independent expert panel uses the Preliminary Report to warn of insufficient safeguards, the too-late efforts to regulate AI after harms have occurred, and the ways in which AI is engineered through day-to-day design and deployment choices. The Preliminary Report looks to initiate a diagnostic and a compass for scientific initiative, economic shifts, human rights priorities, societal health/education applications and security, and governance conditions for AI futures. For the next phase of the Dialogue, the Scientific Panel calls for options for progressing their work.
Below are pointed recommendations from Research ICT Africa, building on our Just AI work developing AI governance through a justice-centric lens:
The work of the UNGDAI and the UN for AI for All must be able to draw long-term, empirical, and quantitative insights following the trajectory of global digital and AI inclusion capacity. Developing for example, unified, empirical methodologies to measure and compare regional compute density and data sovereignty across disparate economic contexts will be crucial. Multi-year empirical studies are needed to trace how market concentration affects local startup survival, talent retention, and labour conditions. Monitoring the political economy over time i.e. building a common understanding of the extent to which digital inclusion metrics are influenced by digital trade policies, competition regulation, and international investment/funding mechanisms for tech infrastructure over time can direct relevant governance for just and equitable AI futures.
We urgently require standardised, open-access methodologies to measure both sides of the ledger for the full impact of AI deployments. These assessments must go beyond intended developmental benefits to account for hidden trade-offs: environmental footprints, labour market disruptions, public sector over-reliance on proprietary platforms, and the systemic distortion of local markets that often accompanies foreign tech entry.
As automated systems are integrated into essential state infrastructure and social services, citizenship risks becoming conditional on algorithmic consent. What does a technically codified and legally robust Right to Opt-Out look like and how do we ensure that individuals and communities including children can refuse or contest automated decision-making? In actionable terms, this requires the development of alternative technical and legal architectures that guarantee access to public services and civil liberties even when a user chooses to detach from automated systems.
Contemporary AI governance scholars point to how the global AI supply chain relies on an invisible shadow economy of data annotators and content moderators working in precarious and often desperate conditions. Our focus in the short term must include inscribing the transnational labour standards and statutory mechanisms that extend fundamental and specific employment protections to these workers. This begins with establishing the key liability provisions for employers and ensuring that the “algorithmic workforce” is afforded at the very least the same dignity and rights as any other labour sector. The commitment must be to dismantling the invisibility of the shadow economy, actively monitoring and driving down the incidence of such precarious work conditions.
Finally, there is a pressing need to establish integrated trade-informed AI and data governance frameworks. As an example, the African Continental Free Trade Area (AfCFTA), specifically the Digital Trade Protocol can be leveraged to address cross-border data and AI governance in an integrated framework. This can harmonise emerging technology standards, and embed regional rights safeguards for AI systems in regions and trading blocks embedding safeguards in a transactional trade reality. Such governance frameworks must safeguard citizen privacy, secure data sovereignty, and address regulatory ambiguity across regional blocs, lowering trade barriers and entry to new markets for local firms on the one hand and scaling responsible AI and other digital innovation in parallel.
From here
The task before us is to ensure the UNGDAI is a real apparatus for structural redistribution, rather than the surveyor of the current digital contest. Collectively, we have the opportunity to deliver AI governance commitments backed by empirical evidence, is appropriately sequenced, measurable, and relevant.
As Research ICT Africa, as we look toward the second session in New York in 2027 and continuing to support the evidence machinery informing contemporary digital governance. It is a pivotal moment, and the provocation must be a just and equitable AI future for All.