Cypress Charitable Trust

The success of the first Global Dialogue on AI Governance will rely on moving beyond high-level rhetoric toward actionable, inclusive multilateralism. A successful outcome would be defined by four key achievements: Bridging Divide: Success requires concrete commitments to capacity-building, including shared high-performance computing resources and open-source models for developing nations. True progress means ensuring AI benefits are not concentrated in a few technologically advanced countries. Interoperability: Rather than creating a single global law, success would involve aligning disparate national regulations. Establishing "interoperable approaches" allows for safe, cross-border AI innovation while maintaining local oversight. Human Rights: A successful dialogue must formalize a global consensus that AI deployment must comply with international law. This includes mandatory transparency for "black box" algorithms, robust human oversight, and protections against abusive surveillance. Science Integration: The inaugural session must effectively integrate the first report from the Independent International Scientific Panel on AI. Success means policymakers actually use these evidence-based assessments to anticipate emerging technical risks. Ultimately, the Dialogue succeeds if it establishes the United Nations as the "universal home" for AI cooperation, ensuring every country—regardless of wealth—has a seat at the table to govern the transformation before it governs them.

• Safe, secure and trustworthy AI
• Protection and promotion of human rights
• Interoperability of governance approaches
• Transparency, accountability, and human oversight

These four pillars represent the essential balance between innovation and protection. Here is why they are indispensable: Safety & Trustworthiness: This is the foundation of adoption. If AI systems are unpredictable, biased, or prone to "hallucinations," the public and industries will refuse to use them. Without safety, AI becomes a liability rather than a tool for progress. Interoperability of Governance: AI is borderless by nature. If every country creates conflicting laws, it creates a "splinternet" that stifles global research and trade. Interoperability ensures that a startup in one country can scale globally without navigating 190 different rulebooks. Human Rights: This is the moral compass. AI has the power to automate discrimination or mass surveillance. Explicitly protecting human rights ensures that technology serves human dignity and freedom rather than being used as a tool for digital authoritarianism. Transparency, Accountability, & Oversight: This addresses the "Black Box" problem. When an AI makes a life-altering decision (like a medical diagnosis or a loan approval), we must know why. Oversight ensures that humans, not algorithms, have the final say and that there is a legal path for recourse when things go wrong. Together, these pillars ensure AI is reliable, globally scalable, ethical, and controllable.

N/A

In the context of Hong Kong's financial sector in 2026, one critical governance gap is framentation gap (interoperability) despite recent regulatory progress. Addressing these is essential for maintaining the city's status as a global financial hub while adopting advanced AI. The Issue: There is no central, overarching AI law in Hong Kong. Instead, guidance is split across the HKMA (banking), HKIA (Insurance), SFC (securities), and the PCPD (privacy). The Challenge: Financial institutions (FIs) face "regulatory overload," making it difficult to maintain a unified compliance strategy across different business units. Gap to Address: Developing a more unified cross-sectoral framework that aligns local circulars with international standards like the EU AI Act to ensure Hong Kong remains a "universal home" for AI finance.

1. A "Universal Home" for Inclusivity: It is the first platform where all 193 UN Member States—regardless of their level of technological advancement—sit at the same table. This ensures that global AI standards are not just shaped by tech leaders but reflect the priorities of the Global South. 2. Tracking & Follow-up: It serves as the primary mechanism for implementing the Global Digital Compact, turning high-level promises about capacity-building and human rights into a recurring, trackable international process.

Its unique strength of AI Dialogue lies in interoperability, aligning disparate national laws to prevent a fractured global market. By integrating the International Scientific Panel's data, it bridges the gap between technical risk and political action, ensuring AI serves humanity collectively rather than major techfin or finteach working in silos.

Stakeholders should operate through a multi-stakeholder hub: Governments define red-lines, Tech Giants provide compute/open-source access for the Global South, and Civil Society audits for bias. For the format, I recommend a "Tracks & Plenary" structure: Technical Track: Integrates Scientific Panel reports into policy. Regional Track: Allows hubs like Hong Kong/GBA to present localized governance models. Action Track: Facilitates public-private "resource-sharing" agreements. The Dialogue should use hybrid "Agile Working Groups" that meet year-round, ensuring the plenary resolution is for final adoption, not just lip service debate.

Within civil society, grassroots advocacy groups focusing on digital labor rights, algorithmic bias, and privacy for refugees are often overshadowed by well-funded tech lobbies. To include them, the Dialogue should: Fund Participation: Establish a "Global North-to-South" travel fund for civil society delegates. Regional Consultations: Hold mandatory pre-sessions in Africa, SE Asia, and Latin America to capture local ethical nuances. Formal Power: Grant NGOs and labor unions "Consultative Status" with the right to provide official rebuttals to technical panel reports.

From a Hong Kong civil society perspective, engagement must move beyond static speeches toward "Policy Hackathons" (Gov-a-thons). These sessions would pair policymakers with local NGOs and data activists to co-draft specific "Sandboxes" addressing regional issues like AI-driven credit bias or gig-worker rights. Additionally, "Citizen Juries" comprised of marginalized users—such as tech-displaced labor or digital privacy advocates—should hold public Q&A sessions with AI developers. Using hybrid immersive platforms (VR), Hong Kong's digital rights groups can participate in gloablly central debates in real-time, ensuring "co-creation" that is technologically agile and human-centric.

From a Hong Kong civil society perspective, effective governance thrives on transparency and inclusive testing. Key examples include: GenAI Sandbox: This allows financial institutions to test AI in a "controlled environment," ensuring consumer protections are integrated before public release. Privacy-Enhancing Technologies (PETs): Platforms utilizing federated learning allow data analysis without compromising personal data sovereignty-a concrete solution for cross-border privacy. Ethical Auditing Frameworks: Adopting "Human-in-the-Loop" mandates for high-stakes financial decisions ensures accountability. Community Data Trusts: These allow groups to collectively manage and license their data for AI training, preventing exploitation and promoting fair compensation