Monash University, Indonesia

A successful first Global Dialogue on AI Governance in Geneva would transition from broad normative commitments to concrete, operational frameworks. Based on the official mandates and current governance landscape, success hinges on several key outcomes. First, the Dialogue must establish genuine, layered interoperability rather than forced global convergence. Success involves creating a coherent system that connects fragmented regional AI initiatives without duplicating efforts or overriding localized risk thresholds. Second, a successful outcome requires tangible mechanisms to bridge the digital divide. As Secretary-General António Guterres has emphasized, avoiding a world of AI "haves and have-nots" demands actionable capacity-building and digital foundations for the Global South. This means establishing formal channels for developing nations to integrate their culturally specific requirements into the global architecture. Third, the Dialogue should define enforceable technical standards under its mandate for "safe, secure, and trustworthy AI". Rather than merely naming ethical principles, success would involve a commitment to verifiable accuracy guarantees for high-stakes AI deployments, moving trustworthiness from a declaration to a demonstrable property. Finally, success depends on the Dialogue functioning as a genuine multistakeholder platform rather than a traditional UN plenary session. Ensuring that civil society, academia, and the technical community have a meaningful role is crucial for embedding human rights, transparency, and robust human oversight into international AI law. Ultimately, a successful 2026 summit will leave Geneva with a concrete, testable blueprint to evaluate at the subsequent Dialogue.

Interoperability of governance approaches;Safe, secure and trustworthy AI

Establishing verifiable safety is the absolute prerequisite for any functional global governance. The current international landscape has successfully established shared normative principles-such as transparency, reliability, and human oversight, but it currently lacks the rigorous technical mechanisms necessary to empirically enforce them. The urgent priority is addressing this gap by transitioning governance from declared trustworthiness to demonstrable trustworthiness. High-stakes AI deployments in areas like healthcare diagnostics, public safety, and biometric authentication frequently make consequential decisions without providing calibrated estimates of their own uncertainty. This creates severe, asymmetric risks that disproportionately impact vulnerable populations in the Global South, as vividly demonstrated by the catastrophic welfare exclusions linked to high biometric authentication failure rates in systems like India's Aadhaar. Active engagement in this area means legally mandating technical primitives like conformal prediction, which provides distribution-free, mathematically verifiable coverage guarantees for predictive models. This approach allows regulators and auditors to independently verify a system's reliability using real-world data without necessarily requiring access to proprietary model internals. Ultimately, we must prioritize Cluster 4(a) because it serves as the vital foundation for the interoperability of governance approaches (Cluster 4d). It is impossible to safely interoperate diverse regional frameworks or successfully adapt global models to distinct cultural contexts unless the underlying AI systems first possess the mathematical capacity to testably guarantee their own accuracy and accurately signal their operational failure modes. Securing this technical safety baseline is our most pressing mandate.

In Indonesia, the gap between declarative AI ethics and verifiable technical requirements presents profound challenges and unique opportunities as the nation accelerates its National AI Strategy (Stranas KA). The most significant challenge is the combination of nascent "soft-law" regulations and vulnerable digital infrastructure. Currently, Indonesia relies on voluntary AI ethics guidelines, such as Kominfo Circular No. 9/2023, while binding legislation remains under cross-ministry discussion. The catastrophic June 2024 ransomware attack on the Temporary National Data Center (PDNS), which paralyzed hundreds of public services, vividly demonstrated the severe consequences of inadequate digital governance. Deploying opaque, uncalibrated AI systems, particularly for public administration or disaster management, without mathematically verifiable guarantees exponentially increases the risk of systemic algorithmic failure in these high-stakes environments. Furthermore, the proliferation of AI-generated deepfakes during the 2024 elections highlighted the urgent need for robust technical verification to protect social cohesion from manipulative synthetic media. This regulatory transition offers a massive opportunity. As Indonesia drafts its forthcoming Presidential Regulation on AI, it can leapfrog basic compliance checklists by legally mandating verifiable uncertainty quantification for all high-risk applications. Regionally, Indonesia is positioned to leverage the newly adopted ASEAN Responsible AI Roadmap 2025-2030 to champion a layered governance architecture. By embedding strict technical auditing requirements and localized cultural adaptation directly into national and ASEAN frameworks, Indonesia can ensure that deployed AI systems are not just rhetorically "ethical," but empirically safe, highly culturally relevant, and technically accountable.

The Global Dialogue on AI Governance serves as a transformative milestone by providing a universal, UN-mandated home for international cooperation on artificial intelligence. Its primary role is to act as a "dialogue of dialogues," strategically designed to connect, complement, and bring coherence to the fragmented landscape of existing AI regulations. Rather than duplicating efforts, it aims to foster genuine interoperability among diverse regional frameworks without forcing monolithic convergence. First, the Dialogue ensures inclusive global representation. For the first time, developing countries and regions in the Global South have a formal, institutionalized seat at the table to share best practices and highlight their specific cultural and socioeconomic needs. This is vital for closing the digital divide and ensuring that AI deployment actively contributes to the Sustainable Development Goals. Second, it champions a genuine multistakeholder approach. By structurally integrating voices from civil society, academia, and the technical community alongside member states, the Dialogue moves beyond traditional plenary sessions. This broad engagement is critical for translating broad normative values into concrete, technical requirements, such as establishing verifiable accuracy thresholds for high-stakes AI systems to protect human rights and ensure algorithmic accountability. Finally, by convening alongside the ITU AI for Good Global Summit in Geneva, the Dialogue functionally bridges the gap between policy design and technical development. It ensures that international governance frameworks do not trail behind, but rather evolve concurrently with rapid technological innovation. Ultimately, the Dialogue transitions international cooperation from isolated regulatory efforts into a cohesive, interoperable, and globally inclusive architecture.

To effectively advance global AI governance, the AI Dialogue must connect with and build upon a polycentric landscape of existing mechanisms to prevent further regulatory fragmentation. Key regional initiatives include the ASEAN Responsible AI Roadmap 2025-2030, which actively shapes national policies and interoperability across Southeast Asia, and the African Union's Continental AI Strategy, which provides a unified framework for data governance and risk mitigation across Africa. On an institutional level, the Dialogue should integrate the normative groundwork of the UNESCO Recommendation on the Ethics of Artificial Intelligence and leverage technical standard-setting efforts by the ITU, whose AI for Good Global Summit is strategically co-located with the 2026 Dialogue in Geneva. It must also interlock with binding or certifiable regimes, such as the EU AI Act and ISO/IEC management standards. The primary added value of the Global Dialogue is its mandate to act as a "dialogue of dialogues". Rather than discarding or competing with existing frameworks, the Dialogue serves as a universal connective tissue. It provides a unique, UN-backed platform to facilitate layered composability, allowing diverse regional guidelines to communicate with a global baseline. Furthermore, the Dialogue brings unprecedented inclusivity. By giving every country a seat at the table, it provides developing nations with a formal, institutionalized mechanism to ensure their specific cultural contexts, socioeconomic priorities, and localized algorithmic risks are actively integrated into the broader international AI architecture.

Stakeholders, including civil society, academia, the private sector, and the technical community, can contribute to the Global Dialogue by participating in virtual consultations, submitting written inputs, and sharing technical best practices. Their active engagement ensures that AI governance is grounded in diverse cultural contexts, human rights, and empirical realities, moving the process beyond purely state-centric negotiations. To maximize its impact, the Dialogue should fully embrace its proposed "Dialogue of Dialogues" mandate. Rather than attempting to centralize global AI control, the format should be structured to connect and synergize existing regulatory initiatives. We recommend the following structural elements: - Inclusive Plenaries: The Dialogue should move beyond traditional UN plenary formats by ensuring civil society and the technical community have meaningful, interactive participation alongside Member States in high-level segments. - Thematic Breakout Discussions: Dedicated sessions focused on mandated clusters, such as bridging the AI divide, establishing verifiable safety, and promoting human rights, will allow technical experts and policymakers to transition from broad principles to operational specifics. - Moderated Roundtables: Facilitated, interactive exchanges between governments and diverse stakeholders are essential to identify areas of regulatory convergence and mutual learning. - Technical Integration: By convening back-to-back with the ITU AI for Good Global Summit, the Dialogue's structure should actively integrate technical insights, ensuring policy design evolves in tandem with technological capabilities.