GlobalTrust One™ / MurMax® Capital

A successful first Global Dialogue would produce three concrete outcomes. First, a shared recognition that AI governance must address the access layer, not only the application layer. Current frameworks focus on what AI does once deployed. They do not sufficiently address who can access AI-enabled systems in the first place. For the 1.4 billion unbanked adults globally, AI-powered financial, governmental, and social services remain unreachable without an identity infrastructure that meets them where they are. A successful Dialogue names this gap explicitly. Second, a commitment to minimum auditability standards for AI systems operating in high-stakes environments — including financial access, benefits disbursement, healthcare eligibility, and law enforcement. These standards must be required at the infrastructure level, not treated as optional overlays. Human oversight and explainability must be built in from inception, particularly for populations that have historically experienced exclusion or exploitation by institutional systems. Third, a formal mechanism for integrating mission-aligned private sector infrastructure providers into global AI implementation programs. International frameworks establish principles. Small, specialized organizations operating at the intersection of technology, compliance, and community access build the infrastructure that makes those principles reachable for individual human beings. The Dialogue should establish a pathway for recognizing and engaging these organizations as implementation partners, not merely as stakeholders. Success is not a communiqué. Success is a Dialogue whose outputs can be traced to changed conditions for the people most at risk of being permanently excluded from the AI economy.

• Safe, secure and trustworthy AI
• Protection and promotion of human rights
• Transparency, accountability, and human oversight
• AI capacity-building

GlobalTrust Oneâ"¢ operates identity governance and financial access infrastructure for underserved and unbanked communities in the United States. Our selection of these four thematic areas reflects both our operational experience and the structural gaps we encounter daily. Safe, secure, and trustworthy AI: Trust is not abstract for our user population. It is earned through every access event being logged, every decision being explainable, and every process having a human escalation pathway. Our platform, MurMax Sentinel™, implements continuous compliance monitoring and auditability as core architecture - not as add-ons. We selected this theme because trustworthiness must be built into the infrastructure layer before it can be meaningful at the policy layer. AI capacity-building: The communities GT1 serves - unbanked adults, non-standard KYC applicants, underserved small businesses - lack not only access to AI-enabled services but also the foundational digital identity infrastructure required to enter those services. Capacity-building that does not address this prerequisite condition will not reach the populations most in need. Transparency, accountability, and human oversight: Our access recovery pathway allows users rejected by standard systems to complete onboarding through an alternate compliance route. This process is fully auditable. Every decision is traceable. Human review is available at every stage. We selected this theme because accountability cannot be retrofitted - it must be designed in. Protection and promotion of human rights: Financial access is a precondition for the exercise of economic rights. When AI systems inherit the exclusions of the identity and documentation systems they depend on, they become instruments of structural inequality. GT1 exists to interrupt that inheritance.

The listed themes do not explicitly address what we consider the most urgent cross-cutting issue: the identity infrastructure gap, a prerequisite for all other AI governance objectives. Every thematic area identified, capacity-building, human rights, trustworthy AI, and transparency, assumes that the populations to be served can be seen and reached by the systems designed to serve them. For an estimated 1.4 billion unbanked adults globally, this assumption does not hold. They lack the standardized identity documentation that AI-enabled financial, governmental, and social systems require as entry conditions. When those systems are deployed without addressing this gap, they replicate and deepen existing exclusions at scale and at speed. This is not a data privacy issue. It is not an algorithmic bias issue. It is a foundational infrastructure issue. The people most at risk of harm from poorly governed AI are the same people who cannot access the systems that AI governance frameworks are designed to protect. We recommend that the Dialogue establish a dedicated cross-cutting workstream on digital identity infrastructure for excluded populations - examining what standards, interoperability requirements, and public-private implementation models are needed to ensure that AI governance frameworks can actually reach every person they are designed to protect. A second emerging issue not captured: the governance of AI systems used in federal and governmental procurement and contracting decisions. As AI is increasingly applied to eligibility determinations, vendor selection, and benefits access in government programs, the standards for auditability, human oversight, and appeal mechanisms in these systems require specific attention separate from commercial AI governance frameworks. GlobalTrust Oneâ"¢ operates at the intersection of both gaps. We are prepared to provide operational data, case studies, and technical testimony in support of either workstream.

In the United States, the governance gaps we identified are producing measurable harm in the financial access and government services sectors, and AI is accelerating the pace at which that harm compounds. Our sector serves underserved and unbanked community populations that represent over 5.9 million U.S. households, according to FDIC data. As federal agencies, financial institutions, and social service providers increasingly rely on AI-enabled systems for eligibility determination, identity verification, and benefits disbursement, they inherit the structural exclusions embedded in the data systems on which those AI models were trained. Standard KYC systems built on credit history, fixed-address documentation, and legacy banking relationships systematically reject the populations these programs are designed to serve. AI does not correct this pattern; it executes it faster and at a greater scale. The most significant challenge in our sector is the absence of a governance standard for AI systems used in access decisions, specifically, what auditability requirements, human override mechanisms, and alternative verification pathways must exist when an AI system makes a consequential determination about a person's eligibility for financial access, housing assistance, or government program participation. The most significant opportunity is that the infrastructure to close this gap exists today. GlobalTrust Oneâ"¢ has built and operationalized an identity governance and access recovery platform that allows populations rejected by standard AI-enabled systems to complete verified onboarding through an alternate compliance pathway fully auditable, fully human-reviewable, and interoperable with federal procurement and compliance frameworks. What is missing is not technology. What is missing is a governance framework that requires AI systems operating in high-stakes access decisions to provide this pathway and that recognizes infrastructure providers building it as essential partners in the implementation of inclusive AI commitments.

The AI Dialogue can perform a function that no existing body currently serves: translating global AI governance commitments into implementation-ready standards that reach the people those commitments are designed to protect. International AI governance has produced significant normative output principles, charters, frameworks, and compacts that establish what inclusive and trustworthy AI should look like. What remains underdeveloped is the connective tissue between those principles and the infrastructure decisions made by governments, financial institutions, procurement agencies, and technology providers at the operational level. The AI Dialogue is positioned to close that gap in three ways. First, by establishing minimum interoperability standards for AI systems operating in cross-border contexts, particularly in financial access, humanitarian disbursement, and government services, so that a person's identity and access status can be recognized and honored across jurisdictions without requiring full re-verification at every border. Second, by creating a structured channel for operational intelligence from mission-aligned private sector organizations to inform policy. The organizations closest to implementation, those building identity infrastructure, access recovery systems, and compliance platforms for excluded populations, have evidence that policy bodies cannot generate through consultation alone. The Dialogue should formalize how that evidence enters the governance process. Third, by serving as the authoritative venue where AI governance frameworks are stress-tested against real conditions. The measure of a governance framework is not how it performs for populations that AI systems already serve well. It is how it performs for the 1.4 billion people those systems were not designed to reach. The AI Dialogue has the convening authority and the UN mandate to hold that standard. We encourage you to use both.

Several existing mechanisms represent both models and entry points for the AI Dialogue's work. The Global Digital Compact provides the normative foundation that the AI Dialogue should treat it not as a reference document but as a binding implementation mandate, with the Dialogue serving as the accountability mechanism for GDC commitments on inclusive and trustworthy AI. The Financial Stability Board's work on AI in financial services and the Basel Committee's emerging guidance on algorithmic decision-making in credit and access determinations represent the regulatory infrastructure through which AI governance principles reach financial institutions. The AI Dialogue should establish a formal liaison with both bodies to ensure that governance standards for AI in financial access are consistent globally. UNESCO's Recommendation on the Ethics of AI and its AI Readiness Assessment Methodology (RAM) provides a country-level diagnostic tool that the Dialogue should actively integrate, particularly for identifying where identity infrastructure gaps are creating AI exclusion at the national level. The World Bank Group's IFC and IDA programs represent the primary implementation channels through which AI-enabled financial and development services reach emerging markets. The AI Dialogue should work directly with IFC to establish AI governance standards as a condition of technology-related project funding — ensuring that any AI-enabled system funded through IFC meets minimum auditability, human oversight, and access recovery requirements. The added value the AI Dialogue brings to all of these is legitimacy, universality, and enforcement architecture. Existing mechanisms are sector-specific, regional, or non-binding. The AI Dialogue, established by UN General Assembly resolution and involving all 193 member states, can elevate and harmonize these efforts into a coherent global framework and can do so with the political weight required to make participation and compliance meaningful.

Different stakeholders contribute distinct forms of intelligence to AI governance, and the Dialogue's format should be designed to integrate all of them, rather than simply collect their perspectives. Governments bring legal authority and a mandate for implementation. Their most valuable contribution is when they focus on identifying specific regulatory and procurement gaps that hinder the application of AI governance principles in operational systems. The private sector, particularly mission-aligned infrastructure providers, offers operational data, practical implementation experience, and insights into what works and what fails in practice. Their contributions should be structured as testimonies, specific, verifiable, and recorded rather than as consultations. Civil society contributes to the lived experiences of the populations most affected by AI systems. Their input is most meaningful when presented alongside data, rather than in place of it; it should pair community testimony with verified metrics from organizations that serve those communities. Academia and the technical community contribute research, standards expertise, and long-term analyses. Their insights should be integrated into the Dialogue's working groups at the thematic level, informing the Proposed Themes and Structure document before the sessions, rather than only during them. For the Dialogue's format, we recommend adopting a two-track structure: one track dedicated to intergovernmental discussions among member states, and a parallel implementation track where private sector organizations, civil society, and technical communities present operational evidence and stress-test proposed standards against real-world conditions. The two tracks should formally exchange outputs; implementation evidence should inform normative decisions, and normative decisions should be evaluated against implementation evidence before adoption. This structure prevents the Dialogue from becoming a forum that produces frameworks nobody implements, ensuring that the populations most affected by AI governance gaps are represented by concrete evidence rather than mere aspirations.

The most critically underrepresented voices in global AI governance discussions are the people whom AI systems most consequentially affect and who are simultaneously least positioned to participate in the forums where those systems are governed. Unbanked and underserved populations in both the Global South and in wealthy nations, including the United States, where over 5.9 million households remain unbanked, are almost absent from AI governance conversations. They are present as subjects of policy concern but absent as contributors to policy formation. This is not a failure of intention. It is a structural problem: participation in international governance forums requires institutional affiliation, internet access, language access, and the discretionary time that economic precarity denies. Indigenous communities, rural populations in low-connectivity environments, informal economy workers, and stateless or documentation-poor individuals are similarly absent, and these are precisely the populations for whom AI-enabled identity verification, benefits access, and financial services carry the highest stakes. To include them, we recommend three mechanisms. First, establish a formal civil society bridge program that funds and credentials organizations already working with underrepresented communities to participate in the Dialogue as authorized representatives, with full speaking and submission rights. Second, require that all written submissions from private sector and government actors include a statement of how their proposed governance approach specifically addresses populations without standard documentation or digital access. This creates accountability for inclusion at the submission level. Third, partner with organizations operating identity and financial access infrastructure for excluded populations, including GT1, to collect and submit verified data on AI system performance for these communities, creating an evidence base that does not currently exist in global governance discussions.

The AI Dialogue's July 2026 session should introduce three engagement formats that move beyond the panel-and-statement model that characterizes most international governance forums. First, implement stress tests. Each proposed governance standard or thematic recommendation should be evaluated in real time against a specific use case presented by an operational organization. A standard for AI auditability in financial access, for example, should be immediately stress-tested by an organization that operates such a system, identifying within the session itself where the standard holds, where it fails, and what would need to change for it to work in practice. This converts abstract frameworks into actionable guidance within the session rather than months later. Second, evidence-paired testimony. Rather than allocating separate time to government statements and civil society interventions, pair each thematic presentation with a matched evidence brief from a private sector or civil society organization that can demonstrate how the governance gap being described manifests in operational reality with data. This format ensures that lived experience and operational evidence inform normative discussions in real time. Third, a structured interoperability working session. Dedicate a session to mapping where existing AI governance frameworks, such as UNESCO's AI Ethics Recommendation, the GDC, the OECD AI Principles, and regional frameworks align, contradict, and create implementation gaps. Make this session's outputs publicly available before the Geneva Dialogue concludes. Practitioners and governments attempting to implement multiple frameworks simultaneously currently have no authoritative guide for resolving conflicts between them. The AI Dialogue is uniquely positioned to produce one. All three formats should be available to remote participants in real time, with simultaneous interpretation in all six UN languages and structured pathways for remote participants to submit questions and evidence that are formally entered into the Dialogue record.

GlobalTrust Oneâ"¢ exemplifies effective AI governance with our innovative Access Recovery Architecture, seamlessly integrated into our identity verification and onboarding framework. When our AI system encounters a user with non-standard documentation, insufficient credit history, or incomplete Know Your Customer (KYC) data, it does not deny the user outright. Instead, it flags the rejection, meticulously logs the specific point of failure, and routes the case to a human reviewer who has access to a comprehensive audit trail. Moreover, our system triggers an alternative verification pathway, adapted to the user's individual circumstances, ensuring that no one is left without options. Every action is logged, every decision is justified, and every outcome can be appealed through our documented process. With the capability to handle 80,871 requests at 100% uptime and a median response time of under 49 milliseconds, we demonstrate that auditability, human oversight, and speed can indeed flourish together. The governance principles embodied in this architecture offer vital policy recommendations: 1. No AI system responsible for significant access decisions should be empowered to terminate a process without providing a valid reason, establishing a human review process, and offering alternate routes for users who do not conform to the system's training data. 2. Auditability should be a prerequisite in procurement, not an add-on. Any governmental or institutional program utilizing AI for access decisions must mandate that vendors deliver full decision logs, human override documentation, and access recovery metrics as essential contract conditions. 3. The performance data from these access recovery systems should be publicly available and used to measure the effectiveness of the AI governance framework over time, thus creating a vital feedback loop between implementation and policy refinement. GlobalTrust Oneâ"¢ is committed to transparency and is ready to provide operational data, technical architecture documentation, and case study materials to the Dialogue Secretariat upon request.