01Gov

Success means the Dialogue produces something governments can actually use. Not another declaration of principles that dissolves into diplomatic language by the time it reaches implementation. The first concrete outcome should be a shared framework for evaluating AI readiness at the implementation layer, not just the policy layer. Many countries currently debating AI governance have never deployed an AI system into a live public service. The Dialogue should surface the gap between countries that are regulating AI and countries that are procuring and deploying it, and treat both as legitimate governance challenges requiring different tools. A second meaningful outcome would be establishing a non-Western reference group of practitioners. Not academics or civil society observers, but operators with direct government AI deployment experience in the Global South. These voices are greatly absent from current "global" governance conversations, yet their contexts expose the real stress points: procurement lock-in, data sovereignty, linguistic exclusion, and the reality that most AI systems arrive as foreign-built products with no local accountability mechanism. A third outcome: honest acknowledgment that capacity-building and governance cannot be sequenced. The current assumption is that countries must first govern AI, then build capacity. That is backwards for most of the world. Many governments are already buyers of AI systems governed by frameworks they had no role in designing. The Dialogue should address that asymmetry directly. Success is not consensus. It is the start of a governance conversation that includes the people closest to implementation.

• AI capacity-building
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Open-source software, open data and open AI models
• Interoperability of governance approaches

The most significant gap is the government-as-buyer problem. Every theme in GA Resolution 79/325 assumes AI governance operates at the level of developers, platforms, and regulators. But for most governments in the Global South, the primary relationship with AI is as a buyer, not a builder. When a government procures an AI system from a foreign vendor, it inherits that vendor's governance assumptions, training data, and accountability structures. There is no international mechanism that addresses this. No multilateral instrument helps a government assess whether the AI system it just procured meets any agreed standard, because those standards were built for a different actor entirely. The Dialogue should consider minimum disclosure requirements for AI systems sold to governments: training data provenance, performance on local languages and cultural context, and oversight mechanisms that transfer with the contract rather than stay with the vendor. A related gap is the role of local GovTech operators as a governance layer. 01Gov's deployment of One across UAE government entities demonstrates a model where governments can own and interrogate the AI they use, rather than simply consume it. This model has no recognition in current international frameworks. It should.

Arabic is the fifth most spoken language in the world, with over 400 million speakers. It is also one of the most underserved languages in AI development. While Gulf states including the UAE, Qatar and Saudi Arabia are making meaningful investments in Arabic language AI, the majority of AI systems deployed across Arab governments today were still built primarily on English-language training data. The performance gap is real and measurable, yet largely invisible in global governance conversations. This is not a technical problem waiting for a technical fix. It is a governance gap. There are no international standards requiring AI vendors to disclose performance benchmarks by language before selling to governments. There are no procurement guidelines helping a ministry of health in Cairo or a municipality in Dubai assess whether the AI system they are buying will actually serve their citizens accurately in Arabic. The Dialogue should treat linguistic equity as a governance issue, not a localization afterthought. That means disclosure requirements, performance standards, and a mechanism for non-English-speaking governments to formally flag failures at the international level.

The UAE and GCC region are among the most ambitious AI adopters in the world. The governance gap here is not ambition. It is the absence of frameworks built for governments operating as fast-moving buyers. Four realities from the ground: - Government teams are procuring AI faster than they can evaluate it. 01Gov works inside this gap daily, supporting many UAE federal and local government entities. The demand for operational AI literacy, not just policy, is significant and unmet. - Competing international frameworks create confusion, not alignment. UAE entities navigate EU, US, and local standards simultaneously. Vendors fill that vacuum. Governance ends up embedded in contract terms rather than principles. - Arabic-language performance remains inconsistent across commercial AI systems despite growing regional investment. When a citizen-facing government service underperforms in Arabic, the impact is immediate and human. - The opportunity is also real. The UAE has a strong open data foundation. Connecting that to open AI infrastructure would give governments genuine oversight of what they deploy. The region is not waiting for governance to catch up. It is building anyway. That is both the opportunity and the risk.

The AI Dialogue can do something no existing forum has managed: create a structured space for governments to learn from each other at the implementation level, not just the policy level. That means moving beyond declarations and into structured knowledge exchange between practitioners: what procurement decisions governments are actually making, what is failing in deployment, and what local solutions are working. The Dialogue should establish a permanent mechanism for this, not just a two-day meeting every year. It should also serve as a counterweight to the current dynamic where governance standards are written by a small group of technologically advanced nations and handed down to everyone else. Meaningful international cooperation means co-authorship, not consultation.

Recently in the UAE, the Mohammed Bin Rashid Innovation Fund in the UAE has directly backed the development of AI systems built specifically for government use. 01Gov's One, an agentic AI platform serving more than the UAE government entities, is one concrete result of that model: national innovation finance backing a local operator building AI that governments can own and interrogate, not just consume. This model deserves international recognition. The Dialogue should actively map and connect national innovation mechanisms that are funding locally-built government AI, and establish a framework for sharing what works across regions. That is the added value only a UN-convened platform can bring: the legitimacy to link these efforts into something coherent and transferable.

Reserve seats for practitioners, not just policymakers. The people closest to government AI deployment rarely appear in these forums. Structure sessions around real deployment cases, not position papers. Keep plenary statements short and working group time long.

Two groups are almost entirely absent from global AI governance conversations: The first is government AI procurement officers in the Global South and Middle East. They make consequential AI decisions daily under frameworks they had no role in designing. The second is local GovTech entrepreneurs building AI specifically for government use in their own regions. They understand the cultural, linguistic, and operational realities that global vendors do not. Their experience is directly relevant and consistently overlooked. Both groups need structured pathways into the Dialogue, not just open submission forms that favor well-resourced organizations.

Peer review sessions where governments present live deployment cases and receive structured feedback from other governments. Not panels. Not keynotes. Actual working sessions where practitioners talk to practitioners.

One concrete model worth examining is 01Gov (a UAE-based GovTech company) and its agentic AI platform One, built exclusively for government use and co-developed with UAE government entities through structured pilots and joint testing, supported by the Mohammed Bin Rashid Innovation Fund. Even at this early stage, the model is producing something most procurement-based approaches do not: a genuine working relationship between government teams and the AI they are shaping. That distinction matters for governance. Effective AI accountability at the implementation layer requires governments to have a real relationship with the AI they deploy, not just a contract. Co-development models like this deserve recognition as governance instruments in their own right.