University of Southern Queensland

A successful Global Dialogue on AI Governance must move beyond high-level consensus toward a functional, three-stage lifecycle: 1. Inclusion: Actively integrating the perspectives of academics and diverse stakeholders to prevent 'regulatory capture' by a few dominant players. 2. Identification: Pinpointing specific, high-priority areas for urgent interdisciplinary research. 3. Implementation: Establishing clear, data-driven action points with built-in monitoring mechanisms to track global compliance and safety. Success is not measured by the dialogue itself, but by the robustness of the oversight architecture that remains after the summit concludes

• AI capacity-building
• Interoperability of governance approaches
• Transparency, accountability, and human oversight
• Social, economic, ethical, cultural, linguistic and technical implications of AI

I prioritized a human-centric approach to AI, with a specific focus on inclusivity for the neurodivergent community. Capacity-Building:I believe urgent action is needed to bridge the knowledge gap, ensuring all stakeholders can identify the specific risks AI poses to marginalized groups. Multidimensional Implications: I focus on the social, ethical, and linguistic impacts to ensure AI tools are culturally and cognitively inclusive, reflecting the varied needs of neurodivergent users globally. Interoperability: Global cooperation on governance is vital to create a predictable environment where inclusive tools developed in one region can be effectively utilized and protected in another. Transparency and Oversight: These are non-negotiable for building trust. Active human oversight is the primary defense against algorithmic bias, ensuring that AI enhances human potential rather than creating new barriers to entry in social and economic spheres. Together, these priorities form a framework for AI that is safe, accessible, and accountable to all.

While the current themes are comprehensive, there is a critical need for a dedicated focus on Neurodiversity and Cognitive Inclusion. Neurodivergent populations including those with autism, ADHD, and dyslexia face a unique set of 'dual-use' outcomes in AI development that are often overlooked in broad ethical frameworks. On one hand, AI offers transformative opportunities for assistive technology, tailored education, and workplace support, enabling neurodivergent individuals to thrive in environments not originally built for them. On the other hand, the risks are significant: algorithmic bias in automated hiring can systematically screen out neurodivergent candidates, and AI surveillance tools may misinterpret non-normative behaviors as suspicious or 'abnormal.' Without a specific cross-cutting focus on this community, we risk building 'standardized' AI governance that inadvertently creates new forms of digital and social exclusion. Success in global AI dialogue requires ensuring that 'human-centric' AI actually accounts for the full spectrum of human cognition.

The governance gaps in our sector present both a critical challenge and a unique opportunity for regional leadership. Currently, the most significant opportunity lies in our ability to foster a 'feedback loop' between the technology sector and academia. By involving researchers early, we can identify how AI capacity-building and transparency measures can be tailored to support cognitive diversity. The challenge lies in the technical and ethical implications of deploying systems across different cultural and linguistic contexts without sufficient prior testing. To mitigate this, our country is prioritizing the development of pre-deployment guardrails. These are designed to evaluate risks particularly for neurodivergent individuals before technology is integrated into public services. By treating academia as a strategic partner rather than an afterthought, we are gaining the necessary evidence to create interoperable governance that protects citizens while encouraging innovation. This evidence-based approach is essential for maintaining public trust and ensuring the technology serves the entire community.

The AI Dialogue can fundamentally reshape international cooperation through three key mechanisms: Resource Equity: It provides a channel for developed nations to share technical architectures and governance tools, reducing the barrier to entry for the developing world. Strategic Leapfrogging: By accessing global research databases and 'lessons learned,' developing nations can implement sophisticated AI oversight more rapidly and efficiently. Cultural Calibration: The Dialogue ensures that governance is not a monolithic 'one-size-fits-all' approach. It encourages the integration of regional ethical factors, ensuring that technology serves the specific linguistic and social needs of diverse communities. Through this exchange, the AI Dialogue ensures that the benefits of AI are distributed globally, preventing a new era of technological isolationism and ensuring that every nation has the tools to manage both the opportunities and the risks of this transformative technology.

Existing mechanisms like the IGF provide a foundation for dialogue, but the Global Dialogue on AI Governance must go further by integrating the 'Big Tech' developers (Alphabet, Microsoft, OpenAI, Anthropic) into a formal accountability structure. The significant added value here is the creation of a standardized monitoring mechanism. We can draw inspiration from Australia's regulatory approach to digital platforms, which demonstrated that a country can successfully impose safety standards on global tech giants to protect its citizens. The Dialogue should scale this approach, ensuring that AI systems are audited for their impact on communities particularly vulnerable or neurodivergent groups before and after release. By building this link between global policy and corporate action, the Dialogue transforms high-level ethics into measurable, global compliance.

The success of the AI Dialogue depends on its ability to reach 'grassroots' stakeholders who are often absent from high-level policy discussions. To achieve this, the format must be radically accessible. We should utilize audio-capture technologies and simplified digital interfaces to ensure that neurodivergent individuals and those with differing cognitive needs can contribute without the friction of dense academic text. Furthermore, the Dialogue should partner with trusted intermediaries such as schools, universities, and religious groups who can facilitate discussions within safe, familiar environments for teenagers and vulnerable populations. By deploying surveys in a wide array of languages and formats, the Dialogue can gather a diverse 'dataset of human values.' This bottom-up approach ensures that the resulting governance frameworks are culturally sensitive and ethically robust, reflecting a global consensus rather than a narrow western-centric perspective.

Underrepresented perspectives specifically those of neurodivergent communities, Indigenous peoples, and vulnerable youth are critical for ethical AI, yet they are rarely included in high-level governance. To move beyond tokenism, we must adopt a 'Capacity-First' model of engagement. This involves: Technical Empowerment: Before inviting feedback, the Dialogue must provide accessible, multilingual, and neuro-inclusive training on AI risks and opportunities. Decentralized Dialogue: Moving the conversation out of boardrooms and into community centers and Indigenous territories. Active Feedback Loops: Ensuring that the input from these communities directly influences 'action points' for developers. By treating capacity building as the first step of governance, we ensure that underrepresented communities have the agency to protect their interests and steer the technology toward more inclusive outcomes

Effective engagement must be omnichannel, meeting stakeholders in their natural environments. I recommend the following structure: Leverage Established Online Forums: Utilize the existing user-bases of AI giants like Anthropic to capture real-time feedback from the global developer and user community. Academic & Youth Integration: Formalize partnerships with schools and universities to host 'Dialogue Circles,' ensuring the next generation of stakeholders is included in the policy-making process. Offline Bridge-Building: Dedicated outreach for the elderly and digitally-excluded populations through physical learning centers and community groups. The added value of this format is its redundancy: by capturing data through online tech forums, academic institutions, and physical community centers, the Dialogue avoids the 'filter bubble' effect. It ensures that 'human-centric AI' is informed by the full breadth of human society—from the highly digital to the completely offline.

An exemplary model of proactive governance is Australia's Online Safety Act, which demonstrates the efficacy of establishing clear, enforceable safety standards for global digital platforms. This 'safety-by-design' approach provides a blueprint for AI governance by prioritizing the protection of vulnerable users before harm occurs. To scale this effectively, I propose a Three-Pillar Accountability Framework: 1. Mandatory Academic Pre-Screening: Before policies are codified into law, they must undergo rigorous interdisciplinary research. This ensures that governance is evidence-based and accounts for the complex social and ethical implications of AI. 2. Multi-Stakeholder Vetting: This research must integrate the lived experiences of diverse communities specifically neurodivergent and Indigenous groups to ensure the resulting 'guardrails' are inclusive. 3. The 'Public Listing' Governance Model: Just as companies must meet strict transparency and auditing standards to remain on a stock exchange, AI developers should be subject to a 'Governance Listing.' Given that these entities generate immense revenue from billions of global users, they must be held to a continuous monitoring standard. Compliance with national guidelines should be a prerequisite for market access, ensuring that corporate profit never supersedes public safety.