Mpesa Africa

A successful first Global Dialogue on AI Governance should deliver outcomes that place people, rights, and trust at the centre of AI development and deployment. This includes establishing a shared global baseline for safe, secure, and trustworthy AI, ensuring that systems are reliable, resilient, and designed to minimize harm, particularly in high-impact sectors. Among other key outcomes, there should be an agreement on safeguarding and fostering people's human rights. All AI technology should be created and used in accordance with principles like dignity, non-discrimination, and increased access to opportunities for everyone; special attention should be paid to cases when AI influences access to financial means, employment, and other vital elements. Transparency, accountability, and human supervision over AI systems must also be ensured through clear action-oriented requirements. The users must understand how decisions were reached using AI, and organizations must remain accountable in terms of their ability to provide auditability, rectify mistakes, and interfere with processes if needed. Finally, the Dialogue should promote and encourage responsible use of open-source software, open data, and open AI models. These approaches will foster innovation, inclusiveness, and cooperation; however, it is important to remember about proper protection of data and AI from any possible abuse.

• Safe, secure and trustworthy AI
• Transparency, accountability, and human oversight
• Open-source software, open data and open AI models
• Protection and promotion of human rights

The priorities reflect the need to ensure that AI systems are effective and responsible, especially in high-impact data-driven sectors. Safe, secure and trustworthy AI is key to mitigate risks like data breaches, system manipulation and unintended harm. As AI becomes increasingly embedded in critical services, reliability and resilience will be key to maintaining public trust The protection and promotion of human rights is a top priority. AI systems should be designed and deployed in ways that respect dignity, prevent discrimination and ensure equitable access to opportunities. This is particularly important in cases where AI affects access to financial services, employment and other vital resources. We need transparency, accountability and human oversight to make sure decisions taken by AI are explainable and open to review. Organizations must be able to explain outcomes, provide avenues for redress and retain meaningful human involvement in the decision-making processes. Lastly, open source software, open data and open AI models can drive innovation, inclusivity and collaboration. But this work must be done responsibly, with the right protections in place to prevent abuse, protect sensitive data and ensure the integrity of the system.

Third-party and vendor risk in AI systems is an important cross-cutting issue. Many organizations rely on external providers for AI tools, models and infrastructure, which raises challenges in terms of accountability, data protection and compliance across the value chain. Standards should be clear to define responsibilities between developers, vendors and deployers. Another emerging issue is data governance, particularly concerning data retention, quality and cross-border data flows. Inconsistent practices can create regulatory gaps and open the door to misuse or unauthorized access, especially when dealing with sensitive personal and financial data. In addition, systemic risk is an increasing concern, especially in the financial services industry. The unchecked growth of AI systems without adequate supervision could bring systemic risks, affecting market stability and consumer trust. In the end, we need to connect global AI governance frameworks to the realities of local implementation. Different regions have different levels of technical capacity, regulatory maturity and infrastructure readiness Global frameworks that are adaptable and inclusive will be critical to the effective and equitable governance of AI.

Gaps in AI governance are impacting sectors across Africa, particularly in digital financial services. One of the biggest challenges is the pace at which AI is being adopted without equally developed regulatory frameworks for safety, security and accountability. This can cause risks like data breaches, fraud, and misuse of AI systems. Limited transparency and explainability also make it hard for users and regulators to understand AI's decisions or to challenge outcomes. This is especially important in financial services, where AI can be used to determine access to credit, detect fraud and profile customers. Data privacy, bias and the potential exclusion of vulnerable groups remain human rights concerns. Without sufficient safeguards, AI systems can unintentionally perpetuate inequities, erode trust in digital services, and more. There are important opportunities at the same time. Open source AI and data enable innovation and allow for locally relevant solutions. Current governance gaps also provide an opportunity for African countries to develop context-specific frameworks that balance innovation and accountability. With the right investments in capacity building and governance, AI can help with financial inclusion, economic growth and more resilient digital systems.

I see the AI Dialogue as a neutral, inclusive global forum that can bring together fragmented efforts in AI governance and bolster international cooperation around shared principles and practical action. Its main function should be to facilitate convergence on shared minimum standards for safe, secure and trustworthy AI, while acknowledging different national contexts. This can help to reduce regulatory fragmentation and increase predictability for governments, industry and civil society . The Dialogue can also build trust among stakeholders by transparently sharing best practices, risk assessments, and lessons learned from actual AI deployments. This is particularly critical in high-impact sectors where AI decisions directly affect human rights, livelihoods and access to essential services. It is crucial that the AI Dialogue enable meaningful participation from developing countries, so that global AI governance is not shaped by a few actors alone and reflects diverse perspectives and needs. It can also facilitate cooperation on shared challenges, such as AI safety, accountability mechanisms, and responsible innovation, including governance of open-source models and cross-border data use. What matters in the end is the Dialogue's capacity to translate disparate conversations around the world into sustained cooperation, concrete commitments, and trust among states and stakeholders.

The AI Dialogue should build on existing initiatives like the OECD AI Principles, the UNESCO Recommendation on the Ethics of Artificial Intelligence and ongoing work under the Global Digital Compact. It should also connect with regional frameworks such as the EU AI Act and African Union digital transformation strategies, along with multi-stakeholder efforts from the Global Partnership on AI and the Partnership on AI. By linking these initiatives, the Dialogue can cut down on duplication and help pinpoint areas of agreement and disagreement in AI governance approaches. Its main benefit would be offering an inclusive UN-led space where governments, private sector representatives, academics, and civil society can engage equally. Unlike scattered regional or voluntary efforts, it can support global legitimacy and wider representation, especially for developing countries. It can also help turn high-level principles into practical governance tools, encourage cooperation between regulatory systems, and promote shared approaches to AI safety, transparency, and protecting human rights.

Governments should share their policy views, regulatory experiences, and national priorities, particularly regarding AI safety, accountability, and human rights protections. The private sector can provide technical knowledge, insights on deployment risks, and practical methods for responsible innovation. Civil society should highlight social impacts, equity issues, and rights-based protections. Academia and the technical community can offer research, risk assessments, and help develop standards. International organizations can support coordination and ensure alignment with existing global frameworks. To encourage meaningful participation, the Dialogue should follow a multi-stakeholder format with balanced representation and organized thematic sessions. It should include both high-level policy discussions and technical working groups focused on specific topics like AI safety, transparency, and open-source governance. Hybrid participation, both in-person and virtual, should be maintained to ensure accessibility, especially for stakeholders from developing countries. Pre-consultations and written submissions should be part of the process to allow broader input beyond session attendees.

Several voices are still underrepresented in global AI governance discussions. These include stakeholders from developing countries, especially in Africa, as well as least developed countries and small island developing states. Local communities that are affected by AI deployment, informal sector workers, and marginalized populations often face exclusion. Additionally, youth perspectives, Indigenous communities, and linguistic minority groups are not consistently included, even though they are significantly impacted by AI systems. To ensure these voices are heard, the AI Dialogue should focus on building capacity and offering financial and technical support for participation. This support should include travel assistance, tools for remote participation, and simpler submission formats. The Dialogue should also set aside specific seats or thematic sessions for underrepresented groups. Partnering with regional organizations and civil society networks can help bring grassroots perspectives into global discussions.

The AI Dialogue should be presented in formats beyond conventional panel discussions. Interactive formats that encourage innovative solutions to current and emerging issues should be the key approach. Scenario-based workshops would enable participants to work through real-world AI risks and examine governance responses. 'Policy hackathons' could identify and propose workable solutions to challenges surrounding issues of AI transparency, safety, and accountability. Citizen panels and public consultations would offer broader public involvement, reflecting citizen concerns and expectations in the Dialogue. Such consultations would be particularly valuable through use of digital platforms to facilitate input before, during, and after the Dialogue. Live simulations of AI risks and governance could help connect technical experts and policymakers. In structured breakout groups organized around various topics, in-depth technical and policy level discussions would be possible, leading to inputs that could be reflected in plenary sessions.

A number of current policies and initiatives serve as valuable examples of how effective AI governance and the practical aspects of addressing emerging issues can be applied: The OECD AI Principles. They emphasize trustworthiness by emphasizing the values of human rights, fairness, transparency, accountability, and robustness. These principles are widely accepted globally and serve as an important standard for responsible AI development. The UNESCO Recommendation on the Ethics of AI. The Recommendation is a global normative standard that centers human dignity, fairness and inclusivity across various legal and cultural settings. The EU AI Act. It implements a risk-based regulatory framework which establishes requirements for high-risk AI systems including transparency, documentation, and human oversight. This serves as a practical model for translating governance into action. Multi-stakeholder initiatives such as GPAI provide examples of inter-state and private actor collaboration on AI safety, data governance and responsible innovation. Within private organizations there are many examples of internal governance frameworks ranging from model risk management systems and ethics review committees to requirements on explainability and accountability. Open-source AI and transparency initiatives also play a role by supporting broad assessment, collaboration and safety. Combined, these examples illustrate that the successful implementation of AI governance requires integrating policy, standards and practical tools for ensuring safe, transparent, and human-rights aligned systems.