From principles to practice
The session moved the Dialogue from principles to practice, asking what actually works when countries, regions and institutions build AI governance. The recurring thesis across speakers was interoperability without uniformity: different legal traditions are fine, but shared terminology, standards and evidence are needed to prevent fragmentation. Rather than debate whether to regulate, the panel compared how governance is being built in very different settings, and where those approaches can connect.
National and regional stacks
Egypt’s Hoda Baraka presented a full “stack”: a National AI Strategy, a National Council for AI, the Egyptian Center for Responsible AI, a national AI governance framework and guidelines, AI procurement guidance, an AI regulatory sandbox, and Karnak, Egypt’s open-source Arabic-language model (framed as digital sovereignty). She stressed a risk-based tiered model, matching the weight of oversight to the level of risk.
For the African Union, Commissioner Lerato Dorothy Mataboge described the Continental AI Strategy (2024), which gives 55 member states a common guideline; 16 African countries now have national strategies of their own. The AU is pursuing regional anchors, shared compute, sovereign data and model control, African-language representation, and “patient capital,” and argues for Africa as an architect, not only a consumer, of AI.
Ana María Ibáñez of the Inter-American Development Bank made the case that Latin America and the Caribbean should evolve the institutions it already has, such as data-protection laws and existing agencies, rather than reinvent them, using blended finance and technical assistance. She cited a Brazilian example: the state of Ceará raised judicial productivity by roughly 40% with AI.
Standards, evidence and the EU framework
ISO Secretary-General Sergio Mujica positioned standards as the bridge between regulation and implementation: standards define the “how,” while regulators define the “what.” The goal, he argued, is a single benchmark that travels across borders.
The OECD’s Yoshiki Takeuchi pointed to a mature toolkit already in service: the OECD AI Principles, the OECD.AI Policy Observatory (2,200+ policies across roughly 90 jurisdictions), the Global Partnership on AI, the AI Incidents Monitor, and the Hiroshima AI Process reporting framework, alongside a new AI Policy Toolkit. He flagged an adoption gap that governance will have to reckon with: about 52% of large firms use AI versus 17% of small firms.
Roberto Viola of the European Commission laid out a four-part strategy: an “AI-first” adoption principle, public compute (AI factories and larger gigafactories), the EU AI Act as a proportionate risk-based framework, and international cooperation. He observed convergence from both directions: light-touch jurisdictions are adding frameworks while regulation-first ones are investing more.
Industry and civil-society counterpoints
Lu Zhang of Fusion Fund argued that governance should be technology-informed, attentive to agentic AI, hybrid large and small models, and federated computing, and should use incentives so that responsible AI becomes a competitive advantage rather than only a compliance burden. Jason Pielemeier of the Global Network Initiative urged the room not to start from scratch: build on more than 20 years of internet governance and ground AI governance in international human rights law, putting the global majority “in the kitchen, helping to set the menu.”
A capacity launch: the UN AI Resource Hub
The session closed with the launch of the UN AI Resource Hub, an enhanced one-stop platform for AI capacity-building and fellowships mandated by paragraph 86 of the WSIS+20 outcome document. It was presented by the UN Inter-Agency Working Group on AI (co-led by ITU and UNESCO, drawing on more than 60 UN entities): Tomas Lamanauskas (Deputy Secretary-General, ITU), Gabriela Ramos (Assistant Director-General for Social and Human Sciences, UNESCO), and Robert Opp (Chief Digital Officer, UNDP). The hub is the practical answer to the panel’s shared-resources thesis, turning a common evidence and skills base into something countries can actually draw on.
Why it matters for the SDGs
The session sits squarely on SDG 9 (industry, innovation and infrastructure) and SDG 17 (partnerships, technology, capacity-building and interoperability): the panel’s whole premise was that governance travels through shared standards, finance and evidence rather than isolated national efforts. The inclusion themes from the African Union, Egypt and the IDB, from African-language representation to evolving existing institutions, map to SDG 10. The shared-standards-and-evidence thread is the measurement angle SDGCounting watches: governance only works if impacts can be observed and compared across borders, which is exactly what one standard, one test and one policy observatory are built to enable.
Watch & read
- UN Web TV, recording of the Day-2 plenary sessions (7 July 2026).
- Independent International Scientific Panel on AI, the shared evidence base speakers referenced.
- Full Global Dialogue coverage.
Quotations are lightly edited from an automated (Otter.ai) transcript of the UN Web TV recording and should be read as close paraphrase; names and titles were reconciled to public records and reflect roles at the time.