Credal.ai Corporation

Success for this first Dialogue won't come from a sweeping declaration or symbolic communiqué. It will come from establishing the right architecture — one that makes governed AI adoption the path of least resistance for every institution that matters to human welfare. I say this from experience. At Credal, we've worked with hospitals, financial institutions, government agencies, and global regulators — organizations with immense AI potential, but that cannot move without ironclad governance. They are not waiting for better models. They are waiting for frameworks and infrastructure that let them act responsibly. The Global Dialogue has a unique opportunity to be the convening force that unlocks this. Three concrete outcomes would signal genuine success: First, agreement on a shared governance vocabulary — a common lexicon for concepts like audit logging, access controls, data residency, and human-in-the-loop oversight. Right now, every regulator speaks a different language, which paralyzes enterprises operating across jurisdictions. Second, a commitment to making governance a prerequisite, not an afterthought. The internet was transformative — and we squandered most of that potential for human welfare because no one built the guardrails in early. AI must not repeat that mistake. Global norms should mandate that AI deployments in critical sectors — healthcare, education, government — ship with documented governance controls as a baseline. Third, a call-to-action directed at the private sector: enterprises and AI vendors alike must be held accountable. Governance cannot live in policy documents alone — it must be embedded in the actual systems people use, enforced in real time, and auditable on demand. This Dialogue succeeds if it plants a flag: that frontier intelligence and responsible governance are not in tension — they are inseparable.

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

My selections are anchored in a single conviction: AI's greatest unfulfilled promise lies not in consumer applications, but in the institutions most critical to human welfare - hospitals, governments, financial systems, international bodies. These institutions want to harness AI, but they cannot act without governance infrastructure that matches the stakes of their work. Safe, secure and trustworthy AI is the non-negotiable foundation. Without it, every other priority is theoretical. At Credal, we've seen firsthand that regulated enterprises don't lack appetite for AI - they lack the guardrails that make deployment responsible. Transparency, accountability, and human oversight must be structural requirements, not optional features. Governance that lives only in policy documents is not governance - it must be embedded in the systems themselves, enforceable in real time, and auditable on demand. Interoperability of governance approaches is the most urgent and underappreciated challenge. Enterprises and governments operating across jurisdictions face a paralyzing patchwork of incompatible frameworks. The UN has a unique convening role to build the shared vocabulary and interoperable standards that unlock responsible AI adoption at global scale. Finally, social, economic, ethical and cultural implications reminds us why this work matters. Like the internet before it, AI risks being a transformative technology that fails to deliver transformative outcomes - unless we build the governance architecture now, not after the damage is done.

Yes - and I'd argue it is the most consequential shift in AI that current governance discourse has yet to fully reckon with: the rise of agentic AI. The themes above were largely conceived in an era of generative AI - systems that produce content in response to human prompts. That world already demanded urgent governance attention. But the frontier has moved. AI systems are increasingly autonomous agents that don't just answer questions - they take actions. They write to databases, send emails, modify records, execute transactions, call other agents, and make sequential decisions across complex workflows - often without a human in the loop at each step. This is categorically different from anything the existing governance lexicon was designed to address. The risks are not just about biased outputs or hallucinated content - they are about consequential, real-world actions taken at machine speed and enterprise scale, across systems that hold some of the most sensitive data on earth. The governance challenge this creates cuts across every theme listed: it demands new thinking on accountability (who is responsible when an agent chain causes harm?), on human oversight (how do you maintain meaningful oversight of systems acting faster than human review cycles?), and on interoperability (how do multi-agent systems operating across vendors, jurisdictions, and data environments respect consistent governance rules?). At Credal, we are already building the infrastructure for governed agentic AI - authenticated agents, authorized actions, full audit trails across every step of every workflow. But infrastructure alone is not enough. The world needs a governance framework purpose-built for the agentic era, before autonomous AI systems become load-bearing pillars of critical infrastructure without the oversight architecture to match. The internet analogy I keep returning to is instructive: we didn't build the security and governance layer into the internet's foundations, and we have spent decades paying for that omission. Agentic AI is that same inflection point - and this Dialogue has a narrow window to get ahead of it.

I'll answer this from where I sit every day: at the intersection of frontier AI and the enterprises most critical to human welfare. Credal works with hospitals, federal government agencies, global financial institutions, and international regulatory bodies. What I observe across all of them is not an AI adoption problem — it is a governance gap problem, and it is acute.

The AI Dialogue arrives at a defining moment — and its role, if seized correctly, is not merely convening. It is architecture. Let me explain what I mean. International cooperation on AI governance has so far produced a proliferation of frameworks, principles, and declarations. The OECD AI Principles, the EU AI Act, the G7 Hiroshima Process, the Bletchley Declaration — each valuable, none sufficient alone. What we have built is an ecosystem of parallel efforts that speak different languages, operate on different timelines, and create as much fragmentation as they resolve. The institutions I work with — global banks, international regulators, health systems operating across borders — feel this fragmentation acutely every day. The AI Dialogue has something none of those efforts had: the convening authority of the United Nations and the participation of the full international community, including the Global South, which has been largely a standard-taker rather than a standard-setter in AI governance to date. I would argue the Dialogue can play three distinct and irreplaceable roles in advancing genuine international cooperation: First, it can be the forum that produces a shared governance lexicon. Before we can have interoperability of governance approaches, we need a common language. What does "audit logging" mean across jurisdictions? What constitutes "meaningful human oversight" in an agentic AI system? What are the minimum standards for "trustworthy AI" in a critical infrastructure context? Second, it can establish the principle that governance is a prerequisite, not a premium. One of the most damaging dynamics in AI today is the implicit assumption that governance is in tension with capability — that regulated, responsible actors must sacrifice speed and power for safety. This is false, and it is dangerous, because it creates a race to the bottom where the least accountable actors set the pace. The Dialogue can plant a global flag: that in sectors critical to human welfare — healthcare, education, public administration, financial infrastructure — AI deployments must ship with documented, enforceable governance controls as a baseline condition. Not a best practice. A requirement. Third, and perhaps most importantly, it can give the private sector a clear signal. Enterprises and AI vendors are not waiting for perfect governance frameworks — they are making irreversible architectural decisions right now, under conditions of regulatory uncertainty. The Dialogue does not need to solve every governance question to be transformative. It needs to provide enough signal clarity that responsible builders can design governance in from the ground up, rather than retrofitting it after the fact. The internet generation did not get that signal early enough. AI governance does not have to repeat that mistake.

I'd like to further discuss all of this as a speaker at the Dialogue.

I'd like to further discuss all of this as a speaker at the Dialogue.

I'd like to further discuss all of this as a speaker at the Dialogue.

I'd like to further discuss all of this as a speaker at the Dialogue.

I'd like to further discuss all of this as a speaker at the Dialogue.