NeuraRock Consulting

The first Global Dialogue on AI Governance will be a success if it delivers three foundational outcomes. First, a shared baseline of priorities. With 193 UN Member States at the table alongside private sector, civil society, academia, and the technical community, the Dialogue must produce a documented consensus (however provisional) on the most urgent governance challenges: safety, accountability, and equitable access to AI's benefits. This baseline should be concrete enough to guide national policy and inform subsequent sessions. Second, meaningful inclusion of the Global South and small jurisdictions. Success cannot be measured by participation of the most technologically advanced nations alone. Jurisdictions like Gibraltar, which sits at the intersection of financial regulation, digital innovation, and emerging technology governance, offer distinct regulatory insights that larger economies often overlook. The Dialogue must ensure that governance frameworks reflect the full spectrum of national contexts, not just frontier AI developers. Third, an actionable roadmap for interoperability. The fragmentation of AI governance across the EU AI Act, US NIST framework, UK approach, and numerous emerging national strategies creates compliance complexity and market distortion. A successful Dialogue would launch a structured interoperability workstream enabling mutual recognition, harmonised definitions, and cross-border enforcement cooperation. From my work building AI compliance platforms (GlobalAIHelp.com) and advising financial and legal institutions on AI governance, I have seen firsthand how regulatory fragmentation creates barriers for smaller actors. The July 2026 session should end with a clear mandate, working groups, and a timeline for 2027, not just high-level declarations.

• Safe, secure and trustworthy AI
• AI capacity-building
• Interoperability of governance approaches
• Transparency, accountability, and human oversight

These four priorities reflect the core of what NeuraRock Consulting and GlobalAIHelp.com engage with daily across financial services, legal, and compliance sectors. Safe, secure and trustworthy AI is the foundation of everything. Without common standards for what makes an AI system safe, organisations cannot responsibly deploy AI and regulators cannot meaningfully supervise it. From my work at the Gibraltar Financial Services Commission and in consulting, I have seen how the absence of shared safety definitions creates inconsistent oversight and real risk. AI capacity-building is critical for closing the global divide. Most governance discussions are dominated by jurisdictions with large AI budgets and mature digital infrastructure. Smaller economies need targeted support, practical toolkits, and knowledge transfer so they can participate in governance as equals, not as rule-takers. This is central to the mission of GlobalAIHelp.com. Interoperability of governance approaches is essential for global commerce. Organisations operating across borders face a patchwork of frameworks including the EU AI Act, NIST AI RMF, the UK pro-innovation approach, and emerging national laws. Without interoperability, compliance becomes a barrier to entry for smaller actors and a driver of regulatory arbitrage. Transparency, accountability and human oversight underpin public trust. AI systems deployed in financial services, public administration, and healthcare must be explainable and subject to meaningful human review. As someone who has built and deployed AI solutions for regulated industries, I know that these are not abstract principles but operational requirements that determine whether AI adoption succeeds or fails.

Several critical cross-cutting issues merit dedicated attention in the AI Dialogue framework. First, AI governance in financial services and RegTech. Financial services are among the highest-stakes domains for AI deployment, yet current governance frameworks treat them generically. AI is now used in credit decisioning, fraud detection, AML/CFT screening, and investment advice. Each of these creates specific accountability, explainability, and consumer protection challenges that go beyond general AI safety principles. The Dialogue should consider a dedicated workstream on AI in regulated industries. Second, agentic AI and autonomous systems. The rapid emergence of AI agents that take actions, make decisions, and interact with other systems without direct human instruction creates a governance gap that none of the seven listed themes fully addresses. Questions of legal personhood, liability allocation, and oversight mechanisms for autonomous AI are urgent and require international coordination before deployment outpaces regulation. Third, AI and financial inclusion. AI systems in credit scoring, insurance pricing, and banking services risk encoding and amplifying existing inequalities, particularly for populations in the Global South and smaller jurisdictions. This intersects with human rights but is sufficiently distinct to warrant standalone attention. Fourth, compute governance and resource concentration. The concentration of AI compute capacity in a handful of private entities and a small number of nations creates structural dependencies that undermine sovereignty and equitable governance. Any credible global AI framework must address access to compute as a geopolitical and equity issue. As founder of GlobalAIHelp.com and a practitioner in a small, regulated jurisdiction like Gibraltar, these issues are not theoretical. They shape what governance frameworks actually work in practice.

Gibraltar sits at the intersection of financial regulation, digital innovation and cross‑border service provision, so AI governance gaps are felt early and intensely in our ecosystem. Existing frameworks in Gibraltar and the UK already impose high standards on governance, data protection and systems and controls, but they do not yet provide a shared, operational definition of what makes an AI system "safe", "explainable" or "high‑risk" in regulated use. This creates supervisory uncertainty for both firms and regulators when AI is embedded in credit decisioning, fraud detection, AML/CFT, sanctions screening and regtech tools. The most immediate challenge is fragmentation. Financial institutions in Gibraltar often passport services or serve clients subject to multiple frameworks (EU AI Act, NIST AI RMF, UK pro‑innovation approach and emerging sectoral guidance). In practice, this means duplicative documentation, inconsistent risk categorisation and a bias in favour of large incumbents that can afford bespoke compliance architectures, while smaller actors – including innovative regtech and fintech firms – struggle to keep pace. At the same time, there are clear opportunities. First, interoperability work at UN level could enable small jurisdictions to "plug in" to a globally recognised governance baseline while preserving regulatory autonomy. Second, capacity‑building that is tailored to supervisors and smaller regulated firms would help translate high‑level principles into usable toolkits and examination practices. Third, the Dialogue can help surface good practice from small, agile jurisdictions like Gibraltar – for example in DLT regulation and cross‑border supervisory cooperation – and apply those lessons to AI. From my work with financial institutions and professional services firms, the core need is not more abstract principles but practical, interoperable governance mechanisms that regulators, banks and regtech providers can all implement and audit consistently across borders.

The AI Dialogue can become the operational hub of international cooperation on AI governance: the place where fragmented regional and national efforts are translated into a coherent, interoperable ecosystem. As mandated under the Global Digital Compact, it is uniquely positioned inside the UN system to provide a universal, inclusive home for these discussions, rather than adding yet another siloed process. First, the Dialogue can act as a "translation layer" between existing initiatives – from the EU AI Act and OECD AI Principles to UNESCO's Recommendation on the Ethics of AI and national risk‑management frameworks – by curating shared taxonomies, reference use cases, and model clauses that regulators and industry can actually implement. This would directly reduce compliance friction for cross‑border actors, especially smaller firms and jurisdictions. Second, it can institutionalise a genuinely multi‑stakeholder exchange of supervisory practice. Annual meetings, structured working groups and ongoing virtual consultations can surface lessons from a wide range of regulatory environments, including small financial centres like Gibraltar, and feed those insights into more formal standard‑setting bodies. Third, the Dialogue can provide political momentum and continuity. By convening governments, industry, civil society and the technical community on a recurring basis and tying its outputs to the Independent International Scientific Panel on AI, it can turn high‑level principles into a sequenced roadmap with timelines, milestones and accountability. From my vantage point working with financial institutions and professional services firms, this kind of practical, interoperable cooperation is what determines whether AI governance helps innovation or simply fragments it.

There is already a dense landscape of AI governance initiatives: UNESCO's Global Forum on the Ethics of AI and implementation of its Recommendation, the OECD AI Principles and AI Policy Observatory, the Global Partnership on AI (GPAI), regional regulatory projects such as the EU AI Act, and emerging Council of Europe and G7/G20 processes. Rather than duplicating their work, the AI Dialogue should position itself as an integrating layer that connects these efforts and fills critical gaps. Concretely, the Dialogue could: align terminology and risk categories across frameworks; host a shared repository of guidance, toolkits and impact‑assessment methodologies produced by partners; and provide space for cross‑regional conversations that currently happen only informally at conferences and summits. It can also amplify perspectives from the Global South and smaller jurisdictions that are under‑represented in existing clubs of predominantly high‑income countries. The added value of the Dialogue is threefold. First, universality: as a UN‑anchored process, it can legitimise and socialise good practice beyond the membership of any single organisation. Second, coherence: by mapping convergences and divergences between initiatives, it can help regulators and firms navigate overlapping obligations and avoid conflicting guidance, which is already a material challenge for cross‑border financial services and regtech providers. Third, continuity: through an annual cycle of consultations, reporting and thematic workstreams, the Dialogue can track implementation over time, not just produce one‑off declarations. As someone working at the interface of AI, financial regulation and compliance, I see significant potential for the Dialogue to connect sector‑specific work (e.g., on AI in finance) with horizontal human‑rights and ethics frameworks, and to transform that combined knowledge into practical tools for supervisors, firms and technology providers.

Different stakeholders should contribute in ways that reflect both their incentives and their concrete experience. Governments and regulators can bring legal mandates, supervisory practice and impact assessments from high‑risk sectors such as financial services, critical infrastructure and public administration. Industry – especially SMEs, regtech providers and infrastructure players – can contribute detailed system‑level documentation, risk‑management practices and evidence of what is realistically implementable at scale. Civil society, academia and technical communities are essential for surfacing rights impacts, societal risks and technical safety considerations that may not yet be visible in regulatory files. To make this work, the Dialogue should combine plenary political sessions with structured, problem‑focused working groups. Thematic tracks could focus on areas such as financial services, public sector use, foundation models, compute governance and capacity‑building, with balanced representation of Global South and small‑jurisdiction stakeholders in each. Formal written submissions could be complemented by curated "implementation labs" where regulators, firms and civil‑society organisations jointly walk through real‑world use cases, failure modes and supervision challenges. The structure should also support continuity. A standing multi‑stakeholder advisory group could help shape agendas, review outputs and ensure that recommendations are grounded in operational reality rather than purely declarative. For small jurisdictions like Gibraltar, virtual participation and clear channels for submitting regulatory experience (for example, on DLT, fintech and cross‑border supervision) are critical so that they can contribute on equal footing with larger economies.

Several communities remain underrepresented in global AI governance debates: regulators and policymakers from small and micro‑jurisdictions; financial‑inclusion and consumer‑protection advocates working at the frontline of credit, payments and remittances; grassroots civil‑society organisations in the Global South; and professionals in "everyday" regulated sectors such as compliance, supervision and legal services who have to implement AI rules in practice. These groups often lack travel budgets, dedicated policy staff or the bandwidth to follow multiple overlapping global processes. As a result, international norms risk over‑reflecting the perspectives of large technology firms, major powers and well‑resourced NGOs. Yet, for jurisdictions like Gibraltar and comparable financial centres, AI governance will be judged by whether it works for supervisory teams of dozens, not thousands, and for SMEs rather than only for frontier labs. The AI Dialogue can address this by: earmarking seats for small‑state regulators and Global South authorities in each thematic track; providing funded fellowships or travel support for community‑based organisations; and designing asynchronous participation channels (written calls, virtual hearings, structured surveys) that do not require physical presence in Geneva or New York. In addition, partnerships with regional bodies and networks – including those focused on financial regulation, human rights and digital inclusion – can help surface context‑specific insights and case studies. This would ensure that the resulting governance approaches work not only for large AI developers, but also for the smaller financial institutions, regtech firms and public agencies that will ultimately be responsible for implementation

To move beyond set‑piece statements, the AI Dialogue should adopt engagement formats that focus on concrete problems, lived experience and co‑design. One option is "governance sprints": short, facilitated sessions where diverse stakeholders work through a specific scenario (for example, AI‑driven credit scoring in a cross‑border financial‑services context) and jointly map risks, responsibilities and governance tools. Another promising format is supervised "model clinics", where technical experts, regulators and civil‑society representatives examine real or synthetic AI systems – including documentation, evaluation reports and impact assessments – to stress‑test proposed governance approaches. This would make discussions on safety, transparency and accountability far more tangible, especially for sectors like banking and payments where explainability and auditability are non‑negotiable. Hybrid participation should be designed in from the start. Structured virtual townhalls timed for different regions, multilingual online consultations, and moderated thematic forums could enable contributions from practitioners who cannot travel but have critical insights, especially in small jurisdictions and the Global South. Output from these formats should feed directly into the work of formal negotiating or drafting groups, with clear feedback loops so participants can see how their input shapes outcomes. Finally, the Dialogue could experiment with "regulatory sandboxes in conversation": sessions where supervisors and firms from different jurisdictions compare how they tested similar AI use cases under different regimes, extracting lessons for interoperability. As someone working with financial‑services and compliance teams, I see particular value in formats that centre real deployments and supervisory experiences rather than purely conceptual debate.

First, the EU AI Act's risk-based classification and conformity assessment framework. By categorising AI systems into prohibited, high-risk, limited-risk and minimal-risk tiers, and mandating specific documentation, human oversight and post-market monitoring for high-risk uses (including credit scoring and insurance), it creates a clear compliance architecture that regulators and firms can implement. This is already shaping how financial institutions in Gibraltar and the UK approach model governance, data quality and bias testing. Second, the NIST AI Risk Management Framework (AI RMF). Its emphasis on govern, map, measure and manage functions provides a flexible, sector-agnostic structure that can be adapted to specific contexts. In practice, this means firms can integrate AI risk management into existing enterprise risk and compliance frameworks rather than building parallel systems, which is critical for smaller organisations with limited resources. Third, regulatory sandboxes and innovation hubs deployed by the UK Financial Conduct Authority, the Gibraltar Financial Services Commission and comparable bodies. These allow firms to test AI applications in financial services under modified rules with close supervisory engagement, generating real-world evidence on risks and mitigations before full-scale deployment. They are particularly valuable for fintech and regtech start-ups that need clarity on how AI governance rules apply in practice. Fourth, multi-stakeholder governance platforms such as the OECD AI Policy Observatory and the Global Partnership on AI (GPAI), which curate evidence, benchmarks and cross-country comparisons. These reduce information asymmetry and help smaller jurisdictions adopt proven practices rather than reinventing them. From my work with GlobalAIHelp.com and financial services clients, the common thread is that effective governance must be usable: it must fit into existing workflows, be proportionate to organisational scale, and provide clear feedback loops between developers, deployers and supervisors.