AI and Quantum International Hub (AIQUAINT)

A successful first Global Dialogue would be defined not by declarations, but by binding commitments with named owners and measurable timelines. Three outcomes matter most. First, the Dialogue must produce a Co-Chairs' Summary that converts findings into actionable policy options — not consensus language that evaporates into diplomatic archives. Each thematic panel must have a named working group with a six-month mandate, drawing on member states, private sector, civil society, academia, and technical bodies. Second, success requires meaningful structural inclusion of the Global South, not as beneficiaries but as co-architects. Developing nations need AI frameworks designed for their contexts. The Dialogue must produce a concrete developing-country capacity roadmap with measurable targets and funded pathways, backed by ITU and technical bodies. Third, the Dialogue must acknowledge that AI governance is already lagging behind deployment. Civilian harm through surveillance tools, deepfake disinformation, and predictive systems is occurring today, largely outside both military governance and most national regulation. A successful outcome commits to closing this gap specifically , not through a new study cycle, but through immediate interim protections. Structurally, success also means institutionalising written submissions, enabling broader civil society and technical participation beyond those with travel budgets and institutional affiliations. The world is not waiting for perfect governance, it is living with imperfect AI today. Geneva must be the moment ambition becomes action, not another milestone in the long arc of well-intentioned process.

• Safe, secure and trustworthy AI
• AI capacity-building
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Transparency, accountability, and human oversight

Safe, secure and trustworthy AI is foundational because trust is the most critical gap this Dialogue must address. Civilian harm through surveillance tools, deepfake disinformation, and predictive systems that fall outside both military governance and most national regulation is already documented. Without safety as the anchor, governance becomes aspirational rather than protective. AI capacity-building is an existential priority for developing nations. Human oversight frameworks designed for well-resourced institutions collapse in underfunded healthcare, legal aid, and public safety systems - precisely where most of the world lives. Governance without capacity is governance without reach. AIQUAINT's work across countries, including an active initiative to establish AI and quantum education labs, is a step in this direction. Social, economic, ethical, cultural, linguistic and technical implications captures what the script terms "Human Futures" , AI-accelerated job displacement, erosion of authorship and research integrity, and the documented underperformance of foundation models in local languages. Multilingual equity is not a secondary concern; it is a governance failure that directly disadvantages populations relying on public AI services in their native languages. Transparency, accountability, and human oversight ties together the others. Private sector actors must disclose model performance by language, geography, and deployment context. Civil society and academia must be formal co-authors of panel briefs, not consultees. Without enforceable transparency obligations, capacity-building and safety commitments remain unverifiable.

Three significant issues fall between or outside the listed themes and merit dedicated attention. Civilian AI in the Harm Grey Zone is not adequately addressed by existing frameworks. AI-enabled surveillance tools, deepfake disinformation, and algorithmic predictive systems are causing documented harm in civilian contexts, yet they fall outside military governance (which addresses weaponised AI) and outside most national AI regulation (which focuses on high-risk commercial applications). This grey zone is not a gap to be filled later; it is where the most vulnerable populations are currently exposed. Quantum-AI Convergence represents a near-horizon governance challenge that none of the listed themes fully anticipate. As quantum computing accelerates AI capabilities, particularly in cryptography, drug discovery, and financial modelling, governance frameworks built on classical computing assumptions will become structurally inadequate. The talent gap in quantum computing, particularly acute across the Global South, compounds this risk. Early governance attention is needed before capability outpaces regulation, as happened with generative AI. AI and Research Integrity / Epistemic Infrastructure has emerged rapidly as generative AI scales synthetic output at speed. Attribution, authorship, and the integrity of scientific and policy research are eroding in ways that undermine the very evidence base that governance bodies rely upon. This affects not just academic publishing but also the reliability of AI-generated policy analyses, legal briefs, and public health information , with disproportionate impact in regions where institutions lack technical capacity to audit AI-generated content. These three areas share a common feature: they are fast-moving, cross-jurisdictional, and harmful at scale today, not in a projected future. The Dialogue's thematic architecture should create space for these issues either as sub-tracks within existing themes or as an explicit emerging-issues working group.

On safe, secure and trustworthy AI, the Gulf states are deploying AI at scale across smart city infrastructure, public services, financial systems, and healthcare, yet regulatory instruments generally lag deployment velocity, particularly for high‑risk, cross‑sector system. The UAE, Saudi Arabia, and Qatar have published national AI strategies, but enforceable oversight mechanisms remain thin. Globally, the limited and fragmented nature of interoperable safety standards means that high-risk AI systems exported from major AI-producing nations are frequently deployed in Gulf and South Asian markets without localised risk assessments or mandatory incident disclosure. On AI capacity-building, India presents both the sharpest illustration of the gap and the greatest opportunity. With one of the world's largest developer communities and a rapidly expanding AI adoption curve, governance infrastructure has not kept pace. Across the Gulf, AI expertise remains heavily concentrated in expatriate technical talent and large vendors, leaving national institutions with limited capacity to audit or govern the systems they procure. On social, economic, and linguistic implications, AI systems deployed for critical public services across India and the Gulf systematically underperform in Hindi, Tamil, Arabic, and dozens of regional languages — encoding inequity into public infrastructure. Labour displacement is accelerating across Gulf logistics and administration and India's vast IT services sector, without regional social protection frameworks equipped to absorb the transition. On quantum, while still early-stage in deployment, the Gulf and India are both making significant sovereign investments, the UAE's quantum programmes and India's National Quantum Mission signal that quantum-AI convergence is a near-horizon reality for these regions. The governance gap here is anticipatory but urgent: cryptographic infrastructure, financial systems, and critical data held today are already vulnerable to future quantum capability. Establishing governance frameworks now, before capability outpaces regulation, is the precise lesson AI failed to teach us in time.

The AI Dialogue occupies a unique institutional position: it sits within the UN system, which grants it the legitimacy no regional body or industry-led initiative can replicate, yet it is not a treaty body, which gives it the agility to convene across geopolitical fault lines that currently fracture AI governance cooperation. Its most consequential role is to function as a translation layer, converting the fragmented landscape of national AI strategies, regional frameworks, and technical standards into a coherent set of interoperable principles that developing nations can actually adopt, adapt, and enforce. This is not harmonisation for its own sake; it is the precondition for any meaningful global baseline. Specifically, the Dialogue can advance cooperation in three ways. First, it can establish a shared incident disclosure norm , a lightweight, voluntary but politically endorsed mechanism through which member states and deploying entities report AI-related harms. Without shared data, governance remains speculative. Second, it can bridge the standards gap by formally connecting the work of ITU, ISO/IEC JTC1, and national standards bodies into a single interoperability roadmap, with funded pathways for developing-country participation in those processes. Third, it can institutionalise the Global South as a co-author of governance, not a recipient. The Gulf states, India, and emerging African AI ecosystems represent the majority of AI deployment contexts globally — their operational realities must shape the frameworks, not merely respond to them. On quantum-AI convergence, the Dialogue has an early-mover opportunity to initiate a dedicated international working group before capability deployment forces reactive governance — the precise failure mode that defined the generative AI era. International cooperation on quantum governance, while premature for binding instruments, is not premature for foundational dialogue.

The Dialogue's value is not in creating parallel structures, it is in doing what no existing initiative can: converting normative consensus into coordinated, accountable implementation. The OECD AI Principles and UNESCO Recommendation on the Ethics of AI provide the strongest existing normative foundations but lack enforceable implementation pathways, particularly for the Global South. The Dialogue should operationalise these by commissioning contextualised implementation guides co-authored with Gulf, South Asian, and African governance bodies, translating principles into actionable policy instruments. Within the UN system, ITU, UNDP, and UNESCO are advancing AI-related work in parallel without a coordinating mechanism that aligns outputs into a coherent governance architecture. The Dialogue is uniquely placed to serve as that coordination layer, creating accountability linkages and clear ownership of cross-cutting deliverables rather than duplicating mandates. At the sovereign level, the EU AI Act and frameworks such as India's emerging AI governance framework are consequential regulatory developments shaping global AI market behaviour. The Dialogue should map their convergences and divergences to identify where a UN-endorsed interoperability baseline is achievable near-term, giving developing nations a credible reference point that is neither a Brussels export nor a bilateral imposition. The Dialogue must also formally engage AI research and practitioner organisations, including bodies such as AIQUAINT, GAFAI, GCRAI, and AIEOU , as structural partners, not peripheral consultees. These organisations operate at the intersection of research, education, and deployment across the Global South, carrying ground-level implementation knowledge that intergovernmental bodies structurally cannot hold. Their inclusion as co-authors of panel briefs and capacity roadmaps would materially improve both the relevance and the legitimacy of governance outcomes. On Quantum-AI convergence, the Dialogue should ensure governance frameworks being built today incorporate quantum resilience, particularly around cryptographic standards and data protection before deployment realities force reactive governance, repeating the generative AI failure. Engagement with the Open Quantum Institute, which bridges quantum research and global equity, and the IBM Quantum Network, which connects quantum deployment infrastructure across research and industry, would bring valuable technical depth and global reach to the governance process.

Each stakeholder class brings irreplaceable value. Member states must share national AI incident and labour displacement data. Private sector actors must disclose model performance by language, geography, and deployment context. Civil society and academia must serve as formal co-authors of panel briefs, not consultees. Technical bodies including ITU must fund developing-country participation in standards processes. To enable this, the Dialogue should structure participation through named thematic working groups with six-month mandates, accept written submissions from non-accredited stakeholders, and produce a Co-Chairs' Summary with clearly named owners and timelines, converting every recommendation into an accountable deliverable.

The most underrepresented voices are communities directly bearing AI's costs: informal economy workers facing displacement, speakers of low-resource languages, persons with disabilities navigating inaccessible AI systems, and frontline public servants in underfunded healthcare and legal aid systems. Indigenous communities whose data is extracted without consent are almost entirely absent. Inclusion requires more than translation, it requires funded participation pathways, asynchronous and offline engagement mechanisms for low-connectivity contexts, and formal co-authorship roles rather than consultation slots. Regional hubs across Africa, South Asia, and the Gulf should serve as structured entry points, not outreach afterthoughts.

Beyond panel discussions, the Dialogue should pilot structured red-teaming sessions where civil society and technical experts stress-test governance proposals in real time. Asynchronous digital deliberation platforms - accessible in multiple languages and low-bandwidth environments , would extend participation beyond those with travel budgets. Regional pre-Dialogue convenings in the Gulf, South Asia, and Africa should feed consolidated position papers directly into plenary sessions. Scenario-based workshops, where participants navigate real AI deployment failures, would ground abstract governance debates in operational reality. Each format should produce a named output , not a summary, but a draft policy option ready for working group action.

Several concrete models merit attention. The EU AI Act's risk-tiered regulatory architecture offers a replicable classification framework, particularly for high-risk public sector deployments. India's Digital Public Infrastructure approach, embedding governance into technical standards at the stack level , demonstrates how developing nations can build accountable AI infrastructure from the ground up. The UAE's AI regulatory sandbox model enables controlled deployment with mandatory evaluation cycles. At the practitioner level, AIQUAINT's offline AI education labs in partnership with national ministries demonstrate that capacity-building governance can be delivered without connectivity prerequisites. Collectively, these models show that effective governance is contextual, not universal.