Loryi AI

To begin with, we should establish a common and flexible framework for good AI that incorporates considerations around transparency, accountability, safety, and human rights. It should have a global scope while being detailed enough to inform action. Secondly, true success involves a commitment to actioning the ideas that have been developed. We should agree on concrete steps forward; like taking risk-based regulation, publishing model documentation, or developing a separate oversight or audit body. Thirdly, the dialogue should guarantee inclusive representation with a strong voice for the Global South. Ghana and other emerging markets should have a seat at the table so that AI governance incorporates their social, cultural, and economic diversity. If we fail to achieve this, we risk exacerbating global inequality. Fourthly, the dialogue should facilitate capacity-building initiatives that provide developing countries with the support they need to develop their own capacity in AI policy-making, research, and development. Fifthly, true success should also be evident in regional coordination on data governance, AI safety standards, and high-risk system management. We should have a roadmap for continued collaboration rather than just a one-off dialogue. Lastly, the dialogue should create a structure for accountability and continuity; like a standing body or annual peer review mechanism; to monitor progress and update guidelines accordingly. Overall, the measure of success should be moving beyond scattered dialogue to coordinated action that balances innovation with ethics and ensures that we're not leaving any part of the world behind.

• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Open-source software, open data and open AI models

From my point of view, all of these themes are important and impact one another, but the following four are the most pressing and need active engagement: 1. Safe, secure, and trustworthy AI. This is the foundation of any discussion on AI governance. The absence of safety, security, and trustworthiness could lead to negative consequences such as biases, misinformation, and privacy issues. This theme ensures that AI systems are trustworthy and aligned to human values before they are deployed at scale. 2. AI capacity-building. This theme is important because most countries and societies are not yet technically equipped or ready to adopt and regulate AI. Building capacity ensures that the global equity gap is addressed and that developing countries are not only recipients of the benefits of AI but active contributors. 3. The social, economic, ethical, cultural, linguistic, and technical implications of AI. AI has profound implications for jobs, education, culture, and society. This theme ensures that the consequences of AI are well managed to avoid unintended consequences such as job loss, homogenization of cultures, and digital exclusion. This theme ensures that the development of AI is human-centered and context-aware, respecting the diversity of cultures and societies. 4. Open-source software, open data, and open AI models. This theme is important because it promotes collaboration and innovation. It removes barriers to entry, especially for young researchers and innovators, especially in developing countries.

There are also some cross-cutting and emerging issues that are not fully captured by the current themes, but are becoming increasingly important to the effective international governance of AI. There is the issue of AI and geopolitics. This issue has come to the forefront because AI has emerged as a critical asset that influences national security, economic power, and international influence. Then there is the issue of environmental sustainability, which is also critical to the international governance of AI. The training and operation of large-scale AI systems require significant computing resources, resulting in carbon emissions that impact the environment. Another important issue that has not been explicitly mentioned by the current themes is the issue of data governance and sovereignty, which is critical to the international governance of AI. There is also the issue of security and the potential for the misuse of AI, which also needs to be explicitly considered by the international community. Finally, there is the issue of the governance of frontier, general-purpose and Integrity-Censored Algorithmic AI systems , which is also becoming increasingly important as the international community seeks to address the governance of AI systems that are becoming increasingly autonomous. In conclusion, although the current themes provide a solid foundation upon which to establish the international governance of AI, the aforementioned issues are critical to the development of a holistic, flexible, and forward-thinking approach to the international governance of AI that is responsive to the dynamic nature of the field.

For Ghana and the African scene at large, the gaps in AI governance, especially with respect to the development of a safe and trustworthy AI approach, capacity building, weighing socio-economic factors, and openness, is doing a little bit of both: slowing things down a bit while also unlocking some opportunities. One of the main challenges is the lack of adequate regulatory and institutional powers to drive AI development in the right direction. Existing laws and regulations are still evolving; therefore, this is causing a degree of confusion for both local governments and local organizations seeking to develop AI systems. This also poses a risk that AI may be used for negative purposes, automated decisions may become biased, and accountability may become a problem. Another issue that is related to the above challenge is the skills gap that currently exists in the local scene. There is a lack of skilled AI experts, data scientists, and policymakers who have adequate technical knowledge in the field. This slows things down a bit for local innovation and also makes it difficult for local organizations to audit or adapt AI systems to local environments. This also makes local organizations more likely to adopt foreign technologies that may not fully take local languages, cultures, and socio-economic realities into consideration. Lastly, there is also a data gap that exists in the local scene, which also poses a challenge for the development of adequate AI systems that are relevant to the local scene.

The AI Dialogue can play a pivotal role in advancing international cooperation by serving as a neutral, inclusive platform that brings together governments, industry, academia, and civil society to align priorities and coordinate action. First, it can help establish shared norms and principles for AI governance. While many countries are developing their own frameworks, fragmentation risks creating regulatory conflicts and uneven protections. The Dialogue can promote convergence around core values such as safety, accountability, transparency, and human rights, while allowing flexibility for local adaptation. Second, it can facilitate knowledge-sharing and capacity-building, particularly for developing countries. By enabling the exchange of best practices, technical expertise, and policy experiences, the Dialogue can help bridge gaps in readiness and ensure more equitable participation in the global AI ecosystem. Third, the Dialogue can support cross-border coordination on high-risk and emerging AI issues, including AI safety standards, data governance, and the management of advanced AI systems. This is essential because AI systems often operate across jurisdictions, making unilateral approaches insufficient. Fourth, it can act as a catalyst for multi-stakeholder partnerships and joint initiatives, including collaborative research, funding mechanisms, and regional innovation hubs. Such partnerships can accelerate responsible AI development while distributing benefits more broadly. Finally, the Dialogue can contribute to accountability and continuity by establishing follow-up mechanisms; such as working groups, reporting frameworks, or periodic reviews; to track progress and adapt to technological changes. In essence, the AI Dialogue can move global AI governance from isolated national efforts to coordinated international action, ensuring that AI development is both innovative and aligned with shared global interests.

First, the United Nations Educational, Scientific and Cultural Organization (UNESCO) Recommendation on the Ethics of AI offers a globally endorsed normative framework on human rights, transparency, and accountability. Similarly, the Organisation for Economic Co-operation and Development (OECD) AI Principles and the work of the Global Partnership on Artificial Intelligence (GPAI) provide policy guidance and technical collaboration platforms. Second, regional regulatory efforts such as the European Union Artificial Intelligence Act demonstrate practical approaches to risk-based AI governance, while initiatives like the African Union Continental AI Strategy reflect growing regional coordination in Africa. Third, multi-stakeholder and industry-led efforts; such as the Partnership on AI, have advanced research, best practices, and responsible AI deployment across sectors. Added Value of the AI Dialogue The AI Dialogue can add unique value in several ways: Global Coordination: It can connect these fragmented initiatives into a more coherent ecosystem, reducing duplication and aligning standards across regions. Inclusivity: Unlike some existing platforms, it can ensure stronger representation from developing countries, including those in Africa, whose voices are often underrepresented. Policy-to-Action Bridge: It can move beyond principles by encouraging concrete commitments, implementation roadmaps, and measurable outcomes. Cross-Sector Integration: It can bring together policymakers, technologists, and civil society in one forum, fostering holistic solutions. Continuity and Accountability: By establishing follow-up mechanisms, it can track progress and ensure that global AI governance evolves with technological advances.

Different stakeholders can contribute to the AI Dialogue by leveraging their unique roles, expertise, and perspectives: Governments can provide policy leadership, share regulatory experiences, and commit to implementing agreed frameworks. Private sector actors can contribute technical expertise, transparency on AI systems, and best practices for safe deployment. Academia and research institutions can offer evidence-based insights, risk assessments, and independent evaluations. Civil society organizations can represent public interest, advocate for human rights, and highlight societal impacts, especially for vulnerable groups. International and regional bodies, such as the African Union, can help coordinate regional priorities and ensure alignment with broader development goals. Recommendations for Format and Structure Multi-Stakeholder Plenary Sessions: High-level discussions to set priorities, share progress, and build consensus across sectors. Thematic Working Groups: Focused groups aligned with key areas (e.g., AI safety, capacity-building, data governance) to develop actionable recommendations. Regional Consultations: Dedicated sessions to capture context-specific perspectives, particularly from underrepresented regions. Technical and Policy Tracks: Parallel sessions separating deep technical discussions from policy and governance debates, ensuring both depth and accessibility. Action-Oriented Outputs: Each session should produce clear deliverables—guidelines, commitments, or roadmaps—rather than general statements. Continuous Engagement Mechanism: Establish standing committees or annual review cycles to track implementation and adapt to emerging challenges.

Several important voices remain underrepresented in global AI governance discussions, limiting the inclusivity and effectiveness of outcomes. First, communities from the Global South, particularly in regions like Africa, are often excluded or underrepresented. Countries such as Ghana face unique challenges; limited infrastructure, linguistic diversity, and different socio-economic priorities; that are not always reflected in global frameworks. Second, local and indigenous communities are rarely included, despite being significantly affected by data extraction, cultural representation in AI systems, and language exclusion. Their knowledge systems and perspectives are critical for developing culturally sensitive AI. Third, youth and early-career innovators are underrepresented, even though they are both primary users and future developers of AI technologies. Their involvement is essential for forward-looking and sustainable governance. Fourth, small and medium-sized enterprises (SMEs) and grassroots tech communities often lack access to global platforms, despite being key drivers of local innovation and adoption. Fifth, non-technical stakeholders, including educators, social workers, and marginalized groups (e.g., persons with disabilities), are frequently excluded from technical and policy discussions, yet they experience the real-world impacts of AI systems most directly. How They Can Be Included Targeted representation quotas in global forums to ensure geographic and demographic diversity Regional consultations and decentralized dialogues to gather local input Funding and sponsorship programs to support participation from low-resource communities Multilingual engagement strategies to overcome language barriers Youth panels and innovation tracks to amplify emerging voices Partnerships with local organizations to bridge global and community-level engagement

To foster meaningful and dynamic engagement, the AI Dialogue should move beyond traditional panels and adopt interactive, inclusive, and action-oriented formats: 1. Policy Hackathons Short, intensive sessions where mixed teams (policymakers, technologists, civil society) co-create solutions to real governance challenges, such as AI risk classification or data-sharing frameworks. Outputs should be draft policies or prototypes ready for adoption. 2. Scenario-Based Simulations Participants engage in role-playing exercises around emerging risks (e.g., AI misinformation crisis or system failure in healthcare). This helps stakeholders understand trade-offs, coordination gaps, and decision-making under pressure. 3. Multi-Stakeholder Roundtables (Small Group Dialogues) Curated, diverse groups (10–15 participants) enable deeper, more candid discussions than large plenaries, ensuring underrepresented voices are heard and integrated into outcomes. 4. Regional & Virtual Co-Creation Labs Hybrid sessions that connect global participants with local communities in real time. This allows perspectives from regions like Africa, Asia, and Latin America to directly shape discussions without requiring full physical presence. 5. "Reverse Panels" Instead of experts speaking to audiences, community representatives, youth, or civil society actors pose questions and challenges to policymakers and industry leaders, shifting the power dynamic. 6. Live Demonstrations & Use-Case Clinics Organizations present real AI systems, followed by structured critique sessions focusing on ethics, safety, and societal impact. This grounds discussions in practical realities. 7. Commitment Forums ("Action Pledges") Dedicated sessions where stakeholders publicly commit to measurable actions, with timelines and follow-up mechanisms to ensure accountability.

Several policies, practices, and platforms offer concrete and scalable approaches to effective AI governance: 1. Risk-Based Regulatory Frameworks The European Union Artificial Intelligence Act classifies AI systems by risk level (e.g., minimal to high-risk) and applies proportionate obligations such as transparency, human oversight, and strict compliance for high-risk applications. This approach balances innovation with safety. 2. Ethical Standards and Principles The United Nations Educational, Scientific and Cultural Organization (UNESCO) Recommendation on the Ethics of AI provides a global framework grounded in human rights, fairness, and accountability, guiding countries in developing national AI policies. 3. Algorithmic Auditing and Impact Assessments Practices such as Algorithmic Impact Assessments (AIAs) help organizations evaluate risks related to bias, privacy, and societal impact before deployment. These are increasingly used in public sector AI systems to ensure accountability. 4. Model Transparency and Documentation Tools like "model cards" and "datasheets for datasets" promote transparency by documenting how AI systems are built, trained, and evaluated. This improves trust and enables external scrutiny. 5. Multi-Stakeholder Platforms Initiatives such as the Partnership on AI bring together industry, academia, and civil society to develop best practices and conduct research on responsible AI deployment. 6. Open-Source and Collaborative Ecosystems Open AI models and datasets enable broader participation, especially in resource-constrained environments, while fostering innovation and peer review. When combined with safeguards, they can democratize AI development.