Lincoln Alexander School of Law

An outcome that will make the first Global Dialogue on AI Governance a success would be the production of a UN AI Governance Interoperability Observatory, lightly institutionalised within an existing UN entity but with an independent advisory board, that could map regimes, host comparative analysis, and provide a stable knowledge infrastructure for under-resourced regulators.

Interoperability of governance approaches

Interoperability matters now because AI systems do not respect borders, and because regulators increasingly need to cooperate across them. Yet its limits must be stated plainly. Interoperability becomes harmful when it operates as a vehicle for lowering rights protections through equivalence shopping; when it privileges corporate compliance convenience over public-interest pluralism; when it exports dominant regulatory models under the language of technical neutrality; and when it pressures states with limited capacity into premature alignment with frameworks they had no role in shaping. The legitimate aspiration is connective tissue between distinct regimes, not a single regime in disguise. Designing interoperability without these guardrails risks reproducing, in governance terms, the very dependencies that data colonialism scholars have identified in infrastructure.

First, AI governance cannot be reduced to AI safety. The dominant safety-frontier framing centres catastrophic and existential risks while marginalising the ordinary, distributed harms experienced now by welfare claimants, migrants, racialised populations, gig workers, and undersampled communities. Both registers matter, but a governance regime weighted only toward frontier risk will systematically under-protect those whose harms are already material. Second, AI governance is, in substantial measure, infrastructure governance. Compute, cloud, semiconductors, undersea cables, training data, and model access shape the political economy within which any regulatory choice is made. A governance conversation that ignores infrastructure cedes the substance of power to the actors who own it. Third, governance is also procurement. For most states, the operative question is not how to legislate frontier models but how to acquire AI in public services responsibly. Procurement is therefore where rights, transparency, and contestability either become real or remain rhetorical. The most underdeveloped opportunities lie in South-South and triangular cooperation. Smart Africa, ASEAN regulator networks, the Pacific Islands Forum's data sovereignty work, and CARICOM's nascent digital agenda offer scaffolding that, if adequately funded, could host genuinely regional norm-development rather than alignment to external templates. Public-interest AI infrastructure (sovereign or pooled compute, regional language models such as Singapore's SEA-LION and African collectives developing African-language models, and shared data trusts) creates governance leverage that purely regulatory tools cannot. The aspiration should be to expand the set of venues in which standards and norms are made, not merely to widen the consultation window of venues already established elsewhere.

For Africa, three structural features matter most. First, there is a regulatory geography of asymmetry: rule-making is concentrated in jurisdictions that host neither the most affected populations nor the most foundation-model deployment risks. Second, the field is bifurcating between rights-based regulation and innovation promotion, generating pressure on lower-capacity states, especially those in Africa to choose between alignment costs and exclusion from data flows. Third, much governance remains downstream of compute, cloud, and platform infrastructure controlled by a handful of firms abroad, meaning that "governance" in many of our African states resolves into procurement, dependency management, and litigation rather than legislation. The landscape, in short, is less a regime complex than an asymmetric ecology, and any honest account of its trajectory must begin from that recognition.

The credibility of any UN AI governance framework will turn on whether it materially supports the institutions that operate AI rules at the level where harms are felt: data protection authorities, equality bodies, ombudspersons, courts, and the civil society organisations that bring complaints. Aspirational text without resourced enforcement compounds rather than alleviates fragmentation. Also, the governance conversation should resist the temptation to treat AI as exceptional. Most of the rights and remedies required already exist in international human rights law, labour law, refugee law, and consumer protection. The task is application and adaptation, not foundational reinvention.

Example: Black Futures by Design: Advancing Racial Justice in AI Governance: https://rshare.library.torontomu.ca/articles/report/Black_Futures_by_Design_Advancing_Racial_Justice_in_Canada_s_AI_Governance/31310602?file=61787137

Digital consultations, stakeholder-specific community townhalls, and research submissions

African, Black, and other "undersampled minorities."

The indigenous models of a sharing circle format.

"Decoding Transparency - How to Foster Public Trust in Responsible AI Innovation in Law Enforcement" United Nations Interregional Crime and Justice Research Institute and the Centre for Artificial Intelligence and Robotics (April 2026). This is an unprecedented and detailed resource on how to foster public trust in responsible AI innovation in law enforcement. Every law enforcement institution in every country should use this as a foundational document for deploying AI systems responsibly. Link to report: https://unicri.org/Publication-Decoding-Transparency-How-to-Foster-Public-Trust-in-Responsible-AI-Innovation-in-Law-Enforcement Effoduh Jake Okechukwu, "Digital Colonialism and Data Sovereignty in Africa: Human Rights Implications of Big Tech Expansion" Chapter 6 in Rutherford, A. (Ed.). (2026). Corporate Responsibility and Accountability in Business Law: Regulation and Practice in Emerging Sectors (1st ed.). Routledge. https://doi.org/10.4324/9781003683636 Effoduh, Jake Okechukwu "Decolonizing the Governance of Artificial Intelligence in Africa: From Normative Mimicry to Epistemic Sovereignty" Science and Public Policy, 2026, 00, 1-13. Oxford University Press. Fola Adeleke, Jake Effoduh, Kebene Wodajo, Rachel Adams et al "Toward AI Governance Alignment in Africa, Middle East, and Türkiye (AMET) Region" (Responsible AI Governance Report, 2026) Global Center on AI Governance pp 1 - 55. https://www.globalcenter.ai/research/toward-ai-governance-alignment-in-africa-middle-east-and-tuerkiye-amet-region Régis, Catherine, Florian Martin-Bariteau, Jake Okechukwu Effoduh, Rachel Adams, Brunessen Bertrand, Carlos Alfonso Pereira de Souza, Peter Parycek, and Hyesun Melissa Yoon. "Governing with AI: Four Actions to Build a Transformative and Resilient Public Administration." (2026).