Buen Vivir Consulting - Gabriela Uberna

The first Global Dialogue on AI Governance will be consequential only if it moves beyond declarations and produces measurable commitments. Success, from our perspective, means three things. First, the establishment of clear, enforceable indicators for responsible AI deployment — with particular attention to agentic AI systems, which currently represent the highest-risk frontier due to their capacity for autonomous action with limited human intervention. Voluntary frameworks have proven insufficient; what is needed are operationalised standards that companies and governments can be held to account against. Second, a credible path toward convergence of AI ethics policies across member states. Not uniformity — regulatory pluralism has legitimate value — but interoperability: systems that are mutually legible, cross-border enforcement mechanisms, and shared red lines that no commercial pressure can erode. The absence of such convergence today creates regulatory arbitrage that undermines every national framework individually. Third, meaningful inclusion of the Global South in governance architecture. AI's most consequential risks — labour displacement, surveillance, resource extraction — are being exported to contexts with the least institutional capacity to absorb them. A dialogue that produces a framework shaped primarily by technologically advanced nations will lack both legitimacy and durability. A successful first Dialogue will not resolve every tension. But it will have established the procedural foundations — agreed indicators, interoperable standards, inclusive participation — on which durable global AI governance can be built. The alternative is a patchwork of national responses that arrives too slowly to the problems already unfolding.

• Safe, secure and trustworthy AI
• Interoperability of governance approaches
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

The convergence of safety, human rights, transparency, and governance interoperability represents the minimum architecture for responsible AI. We are at an inflection point where AI systems - particularly agentic models operating with high autonomy - create structural risks that individual actors cannot contain alone. Human oversight is precondition. Without meaningful accountability mechanisms, the benefits of AI will accrue asymmetrically, deepening existing social and economic inequities. Resource-intensive AI infrastructure also introduces compounding environmental pressures - from energy consumption to water depletion - that disproportionately burden vulnerable communities. Addressing AI governance in isolation from human rights frameworks and constitutional legal traditions will produce fragmented, unenforceable standards. True progress requires aligning AI development with the Universal Declaration of Human Rights and ensuring that governance approaches across member states are mutually legible and interoperable - not merely compatible on paper.

The intersection of AI and climate accountability remains unaddressed - energy-intensive systems externalise environmental costs invisibly. Equally absent: AI's role in accelerating misinformation at democratic scale.

In EU ESG and AI compliance advisory, regulatory fragmentation creates contradictory obligations for clients. Without interoperable standards, compliance becomes a cost burden rather than a driver of responsible practice.

The AI Dialogue occupies a rare institutional position: it is convened under the legitimacy of the United Nations while remaining structurally open to participation across the full spectrum of member states, civil society, and the private sector. That combination is not replicated elsewhere in the current governance landscape. Its most consequential role is not to produce another framework — the world has frameworks. It is to function as the connective tissue between them: translating commitments made in technically advanced regulatory environments into standards that are legible, implementable, and enforceable across diverse national contexts. In practice, this means the Dialogue should prioritise convergence over comprehensiveness — identifying the minimum viable set of shared principles on which interoperable governance can be built, rather than attempting to resolve every contested question simultaneously. The Dialogue can also serve a critical accountability function. Many existing AI governance initiatives lack review mechanisms with real consequence. By establishing a recurring, structured forum with defined indicators and transparent reporting, the Dialogue creates the conditions for genuine progress tracking — and for naming, without diplomatic evasion, where commitments are not being met. Finally, the Dialogue has a redistributive role to play. Technical governance capacity is not equally distributed. For AI governance to be globally legitimate, it must actively transfer expertise, tools, and decision-making influence to regions currently positioned only as recipients of AI's risks. The Dialogue should institutionalise this — not as a side workstream, but as a design principle embedded in every mechanism it creates.

UNESCO's Recommendation on the Ethics of AI (2021) established the broadest intergovernmental consensus to date on foundational principles. The OECD AI Principles and the Global Partnership on AI (GPAI) have developed technical standards and multistakeholder engagement models with proven implementation track records. The EU AI Act represents the most advanced binding regulatory architecture currently in force, offering a risk-tiered compliance model that other jurisdictions are already referencing. The Hiroshima AI Process and the US AI Safety Institute network have begun operationalising safety evaluation at national level. The AI Dialogue should not duplicate these efforts. Its added value lies precisely in what none of them can deliver individually: universality of participation, UN-level political authority, and the structural capacity to broker agreement between actors who do not share a regional or economic community. Concretely, the Dialogue should function as an alignment layer — mapping existing frameworks for compatibility, identifying contradictions that create regulatory arbitrage, and facilitating mutual recognition agreements between jurisdictions. It should also serve as an amplification mechanism for initiatives that lack reach: many of the most robust AI ethics tools developed by GPAI and UNESCO remain underutilised in the Global South due to capacity constraints, not lack of relevance. The Dialogue's unique contribution, if it is realised, will be to transform a fragmented constellation of national and regional initiatives into a coherent, accountable, and genuinely global governance architecture — one where no actor can opt out of shared responsibility by pointing to the inadequacy of others.

Civil society sets red lines. Industry shares technical realities. Governments commit. Academia evaluates. Structure must enforce parity — no stakeholder group should hold veto power over outcomes.

Global South governments, Indigenous communities, informal workers, and people with disabilities. Include via funded delegations, translated materials, and binding seat quotas — not optional consultations.

Structured adversarial panels — not consensus theatre. Red-team sessions where civil society formally challenges industry claims. Binding citizen assemblies feeding directly into negotiated text.

The EU AI Act's risk-tiered compliance model. Singapore's Model AI Governance Framework. Chile's AI National Policy centering rights. All share one feature: enforceable obligations, not voluntary pledges.