ThetaRay

Success looks like three specific breakthroughs: Uniformity Over Fragmentation: We achieve a global explainability standard for Agentic AI. The Dialogue is a success if it moves past vague ethics and mandates that autonomous systems provide a clear, auditable rationale for every decision. Beyond safety, the reason is to give humans the strategic control necessary to trust these systems at scale. A Sanitary Layer for Global Trade: Success is a commitment to aggressive, secure data reciprocity. We must break the silos that allow criminals to hide in the noise. By establishing a global protocol for sharing anonymized risk data, we can build a sanitary layer that empowers banks in Africa and Latin America to connect seamlessly to global markets without being marginalized by outdated de-risking practices. The Workforce Evolution: We transition from check-the-box talent to strategic decision makers. A successful Dialogue will catalyze educational frameworks that upskill the workforce to manage AI. The goal is to restore capacity, letting AI handle the complexity so that human judgment can focus on high-level risk strategy. If we leave Geneva with a roadmap that treats AI as the new plumbing of global finance rather than a tech experiment, we have succeeded.

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

Our selection reflects the reality that for AI to be a global economic driver, it must first be reliable infrastructure. We prioritize these four areas to transition the industry from experimentation to trusted execution. Transparency, Accountability, and Human Oversight: This is the cornerstone of our Human-in-Control philosophy. As we deploy Agentic AI, systems capable of autonomous investigation, governance must mandate an auditable rationale for every conclusion. We must move beyond black box models to a framework where humans act as strategic supervisors, possessing the transparency needed to tune AI logic in real-time. Interoperability of Governance Approaches: Financial crime is borderless; our rules cannot be fragmented. We advocate for a sanitary layer of shared intelligence through international standards for anonymized data reciprocity. This allows risk patterns to be shared across jurisdictions without compromising privacy, closing the gaps that criminal networks currently exploit. AI Capacity-Building: We are actively bridging the gap in Africa and Latin America. Capacity-building beyond tools, means democratizing the ability of local banks to connect to global markets. By providing turnkey AI infrastructure, we empower institutions in these regions to meet international standards instantly, fostering inclusive growth and sustainable trade corridors. Safe, Secure, and Trustworthy AI: Trust is the plumbing of the financial system. We prioritize safety by replacing static, easily bypassed rules with dynamic, cognitive monitoring. By building systems that learn and adapt, we create a secure environment where legitimate capital flows freely while illicit activity is identified with surgical precision, ensuring the long-term integrity of the global financial grid.

The existing themes cover the fundamental "what," but they miss the critical "how" regarding the speed of modern threats and the underlying infrastructure required to fight them. To make governance effective, two emerging issues must be prioritized: Asymmetric Speed and Real-time Adaptability: Criminal networks operate with total agility, unburdened by legacy systems or borders. Governance often defaults to static, periodic reviews that are obsolete by the time they are implemented. We must address the need for dynamic regulatory sandboxes and real-time simulators. These allow for the continuous tuning of AI controls-what I call Cognitive Monitoring-to match the velocity of illicit activity without disrupting legitimate global trade. The Sanitary Layer of Shared Intelligence: While interoperability is listed, it doesn't go far enough to address the data silo crisis. We face a global intelligence gap because risk data is trapped within national or institutional borders. An emerging priority should be the creation of a global, anonymized sanitary layer for data reciprocity. This would allow institutions in emerging markets, like those in Africa and Latin America, to train their AI on global risk patterns, ensuring they aren't unfairly penalized or de-risked simply due to a lack of local data.

In the financial sector, the governance gap is a velocity gap. While Agentic AI identifies and investigates threats at machine speed, regulatory frameworks remain anchored in manual, point-in-time audits. This mismatch creates friction that disproportionately impacts emerging markets. The Challenges The De-Risking Trap: Governance lacks global standards for data reciprocity. Without shared, anonymized risk intelligence, many banks in Africa and Latin America are "de-risked"—severed from the global financial system—not because they are illicit, but because they lack the data to prove they are safe. Analytic Fatigue: The sector is buried in noise. Legacy systems produce overwhelming false positives, forcing human investigators into low-value data entry. This intelligence gap prevents professionals from focusing on high-level risk strategy. The Opportunities AI as Digital Infrastructure: We can implement AI as a turnkey infrastructure. By adopting Cognitive Monitoring, institutions can modernize instantly, bypassing decades of legacy debt. This allows a bank in an emerging market to meet the same transparency standards as a Tier-1 global bank, leveling the economic playing field. Agentic Oversight: The shift toward Agentic AI—where systems provide an auditable rationale for every conclusion—allows us to restore human capacity. The opportunity is to transition the workforce from check-the-box analysts to strategic supervisors. By closing this gap with interoperable, AI-driven standards, we transform compliance from a barrier into a catalyst for inclusive, borderless trade.

The AI Dialogue should act as the architect for a Global Sanitary Layer of trust. Its most vital role is to move international cooperation from high-level principles to functional interoperability. Specifically, the Dialogue can: Standardize Data Reciprocity: Facilitate a global framework for the exchange of anonymized risk intelligence. By creating a unified protocol for data sharing, the Dialogue can help dismantle the silos that allow financial crime to hide in the friction between jurisdictions. Validate Agentic Oversight Models: Instead of disparate national rules, the Dialogue can establish a global benchmark for Explainable AI. This ensures that as Agentic AI systems investigate and make decisions, they do so against a globally accepted standard of auditable rationale, preventing a race to the bottom in regulatory rigor. Bridge the Inclusion Gap: It can serve as a platform to ensure that the Global South—particularly Africa and Latin America—is not an afterthought. The Dialogue should promote AI as a digital infrastructure that allows these regions to leapfrog legacy compliance debt and connect to the global financial grid with speed and certainty. By focusing on these practical mechanisms, the Dialogue transitions AI governance from a defensive cost-center into a strategic enabler of secure, inclusive global trade.

The AI Dialogue should build upon the foundational work of the FATF (Financial Action Task Force) regarding digital identity and the OECD AI Principles on transparency. It should also connect with regional fintech hubs in emerging markets that are already deploying Cognitive AI to solve local challenges. The Added Value of the AI Dialogue: Moving Beyond Check-the-Box: While FATF provides the "what," the AI Dialogue can provide the "how" through technology. It can bridge the gap between regulatory intent and technical execution by advocating for dynamic, AI-driven monitoring over static, binary checklists. Centralizing Data Ethics for Finance: Existing initiatives often treat AI as a horizontal topic. The AI Dialogue brings unique value by focusing on Vertical AI in high-stakes sectors like finance. It can lead the creation of a global repository for anonymized patterns of legitimacy, allowing AI models to be trained on diverse, global datasets. This prevents the bias inherent in models trained only on Global North data. The Talent Roadmap: Building on UNESCO's work on AI education, the Dialogue can specifically address the upskilling of the financial workforce. It can catalyze a global shift where the compliance professional is no longer a data processor, but a strategic supervisor of AI-driven systems. The Dialogue's ultimate value lies in its ability to synthesize these disparate initiatives into a single, cohesive roadmap for AI as a foundational infrastructure of global trust.

The AI Dialogue must move beyond the summit format to a collaborative lab structure. To ensure it yields actionable governance, I recommend: Public-Private Workstreams: Create dedicated tracks where tech providers, regulators, and NGOs don't just speak, but co-author "Implementation Blueprints." We need practitioners who understand the plumbing of AI to work alongside policymakers to define auditable rationale standards. Sector-Specific Forums: AI governance in finance is fundamentally different from AI in healthcare. The Dialogue should include Vertical AI roundtables where industry-specific risks and infrastructure—such as financial crime and cross-border trade—can be addressed with precision. Technical Evidence Sessions: Allow tech leaders to demonstrate Agentic AI in controlled, simulated environments. Seeing the technology in action—rather than reading a white paper—provides the clarity regulators need to build guardrails that don't stifle innovation.

The most critical underrepresented voices are the regional financial institutions in Africa and Latin America. These entities often lack a seat at the table where Global Standards are set, yet they are the ones most impacted by the "de-risking" that occurs when these standards are too rigid or data-hungry. How to include them: Regional Hub Integration: Connect the Dialogue directly with regional fintech associations and local central banks in the Global South. Equity-Based Participation: Governance shouldn't just focus on "safety" for developed markets; it must focus on access for emerging ones. We must invite local bank CEOs to share how AI infrastructure—not just policy—can help them meet international compliance requirements. The User Perspective: We need to hear from the compliance officers on the ground who are managing legacy debt. Their perspective is vital to understanding how AI can act as a force multiplier to restore human capacity and judgment.

We should replace traditional panels with outcome-oriented formats: The Red Team Simulation: Conduct a live simulation where a regulatory framework is attacked by a hypothetical criminal network. This forces participants to see where static rules fail and where dynamic, AI-driven oversight and Cognitive Monitoring is required. Governance Sprints: Model the Dialogue after tech sprints. Groups are given 48 hours to solve a specific governance gap—such as an "Interoperability Protocol for Anonymized Data"—and present a draft framework to the General Assembly. Reverse Town Halls: Instead of leaders speaking to an audience, let the Human Supervisors (the analysts and investigators who will use this tech) speak to the policymakers. Hearing from the workforce being upskilled provides a grounded reality check on the Human-in-Control standards we are trying to build.

To address the challenges of global financial integrity and economic inclusion, we advocate for and implement three concrete approaches: Cognitive Monitoring and Explainable AI Standards: At ThetaRay, we have replaced legacy, rule-based systems with Cognitive AI. Unlike black box models, this approach ensures every alert comes with a clear, auditable rationale. This is a practice that offers a concrete solution to the transparency gap: it allows human supervisors to understand the why behind a decision, fulfilling the requirement for meaningful human oversight while managing massive data volumes at machine speed. Infrastructure-as-a-Service for Emerging Markets: We have successfully deployed AI platforms as a turnkey infrastructure in Africa and Latin America. By providing local banks with high-level compliance capabilities out of the box, we allow them to bypass decades of legacy tech debt. This practice effectively counters de-risking by giving these institutions the transparency needed to maintain relationships with global correspondent banks, ensuring they remain connected to the global financial grid. The Self-Service Simulator for Dynamic Governance: We utilize a simulator approach that allows for the battle-testing of AI controls in a safe environment. This practice enables institutions to tune their AI logic against emerging criminal patterns before deployment. It serves as a model for dynamic governance, moving away from static annual audits toward a system of continuous, proactive refinement.