PAWA Initiative/Shecode.ai

A successful first Global Dialogue on AI Governance would produce outcomes that are genuinely inclusive of voices from the Global South - particularly Africa, where AI adoption is accelerating but governance frameworks remain nascent. Success would mean: first, the adoption of clear recommendations that acknowledge the digital infrastructure divide, recognizing that meaningful AI governance must address connectivity gaps, energy access limitations, and the lack of localized datasets that currently hinder AI deployment across African communities. Second, the Dialogue should yield actionable commitments to bridge the AI capacity gap - investing in technical education, digital skills, and local AI development ecosystems that empower nations to build, not just consume, AI tools. Third, success requires centering ethical AI principles that reflect diverse cultural, linguistic, and social contexts - not frameworks designed solely for high-income economies. This includes commitments to address gender bias in AI systems, ensuring women and girls are not further marginalized by algorithmic decision-making in health, finance, education, and employment. Fourth, the Dialogue should produce a roadmap for inclusive AI governance architecture that brings civil society, private sector innovators, academia, and marginalized communities into formal governance spaces - not just as observers but as co-creators of standards. Finally, a concrete follow-up mechanism to track implementation, with representation from underrepresented regions, would signal that this Dialogue is more than symbolic - it is the foundation of an equitable AI future.

• Safe, secure and trustworthy AI
• AI capacity-building
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Protection and promotion of human rights

These four themes are interconnected and critical from an African civil society perspective. Safe, secure and trustworthy AI is foundational - without trust, AI adoption will stall or cause harm, particularly in communities with limited recourse mechanisms. AI capacity-building is urgent for Africa: the continent cannot meaningfully participate in AI governance if it lacks the human capital and infrastructure to develop, deploy, and evaluate AI systems locally. Social, economic, ethical, cultural, linguistic and technical implications of AI captures the breadth of challenges we face on the ground - from AI models trained on non-African data producing biased outputs, to cultural erasure through homogenized language models, to economic disruption in informal labor markets. Critically, gender bias in AI systems disproportionately affects women and girls in Africa, reinforcing existing inequalities in hiring tools, credit scoring, and health diagnostics. Finally, protection and promotion of human rights ensures that AI governance is anchored in dignity and non-discrimination - essential where regulatory oversight is nascent and communities are most vulnerable to algorithmic harm. Together, these themes reflect the governance priorities of the 1.4 billion people on the African continent who are both AI's greatest potential beneficiaries and its most underprotected.

Several critical cross-cutting issues remain underrepresented in the current thematic framework. First, AI infrastructure inequality deserves explicit attention. In many African nations, unstable power supply, limited broadband connectivity, and the absence of local data centers make AI deployment not just difficult but inequitable. Governance frameworks that do not address physical and digital infrastructure gaps will remain disconnected from lived realities. Second, gender-responsive AI governance is a cross-cutting gap. Gender bias is not merely an ethical issue - it is an infrastructure, economic, and rights issue simultaneously. AI systems deployed in agriculture, health, and finance in Africa frequently underperform for women due to biased training data and male-dominated design processes. A standalone thematic commitment to gender-responsive AI is warranted. Third, the local context adaptation of AI models needs formal governance attention. Global AI systems often fail to account for indigenous languages, local legal contexts, and culturally specific norms. AI governance must include mechanisms to validate and adapt AI models for local deployment rather than assuming global models are universally appropriate. Fourth, youth and intergenerational equity in AI is an emerging issue: young people in Africa are among the most exposed to AI-driven labor disruption and the least represented in governance processes. Deliberate inclusion of youth voices in formal AI governance structures is essential for long-term legitimacy and relevance.

In Nigeria and across Sub-Saharan Africa, governance gaps in AI are producing compounding challenges with significant real-world consequences. The most immediate challenge is the absence of a localized AI regulatory framework. While Nigeria has introduced a National AI Policy, implementation remains nascent, leaving a vacuum where commercial AI systems operate without meaningful oversight. This gap is acutely felt in sectors like financial technology, healthcare, and agriculture, where AI tools are being rapidly deployed but without mechanisms for accountability or redress when they fail. Infrastructure gaps compound this governance deficit. Unreliable electricity, limited broadband penetration outside urban centers, and prohibitively expensive data costs mean that many communities are subject to AI decisions - in credit scoring, recruitment screening, or government service delivery - without the ability to access, question, or challenge those systems. From a civil society and gender perspective, AI systems trained predominantly on Western, male-skewed datasets routinely misidentify, misclassify, or exclude African women. In the work of PAWA Initiative and Shecode.ai, we observe this in how women-led small businesses are disadvantaged by AI-driven credit risk models and how female professionals are underrepresented in AI-generated content or recommendations. On the opportunity side, Africa's youthful population and rapidly expanding digital economy position the continent as a major beneficiary of well-governed AI. Investments in local AI talent pipelines, open and representative datasets, and inclusive policy design processes could accelerate development outcomes significantly. The AI Dialogue offers a critical moment to ensure governance frameworks create enabling conditions - not barriers - for Africa's AI future.

The AI Dialogue has a unique opportunity to serve as the primary multilateral platform for building consensus on AI governance principles that are binding, inclusive, and enforceable - not merely aspirational. Its most critical role in international cooperation is ensuring that AI governance does not replicate existing global power imbalances, where technology standards are set by a handful of high-income nations and adopted by the rest as non-negotiable defaults. The Dialogue can advance cooperation by establishing a South-South and North-South knowledge exchange architecture on AI governance - enabling countries like Nigeria, Kenya, Rwanda, and Ghana, which are building domestic AI frameworks, to share experiences and receive targeted support. It should also promote common interoperability standards for AI systems across borders, particularly in cross-border digital trade, migration management, and health data systems, where fragmented AI governance creates friction and harm. Critically, the Dialogue can institutionalize a civil society and private sector engagement mechanism that goes beyond symbolic consultation. Africa's tech civil society - organizations working at the intersection of AI, gender, youth, and digital rights - must have a formal seat at the table in designing governance norms that affect their communities. Finally, the Dialogue can play a catalytic role in mobilizing funding for AI governance capacity in the Global South - supporting national regulators, universities, and civil society organizations to build the expertise needed to participate meaningfully in international AI governance, rather than simply receiving and implementing externally-designed frameworks.

Several existing mechanisms provide a strong foundation the AI Dialogue should actively build upon. The ITU's AI for Good platform has cultivated a global network of AI stakeholders and should be formally connected to the Dialogue's outreach and participation infrastructure. UNESCO's Recommendation on the Ethics of AI (2021) provides an already-ratified ethical framework that the Dialogue should reference as a baseline rather than starting from scratch on ethics. The Internet Governance Forum (IGF) and its Dynamic Coalitions on AI and Data Governance have built multi-stakeholder expertise over years and should be designated as a formal input mechanism to the Dialogue. The African Union's Digital Transformation Strategy and the Smart Africa Alliance represent key regional frameworks the Dialogue should explicitly integrate, ensuring that Africa's own governance priorities shape global norms rather than being retrofitted to them post-hoc. The UN Secretary-General's AI Advisory Body report also provides actionable recommendations the Dialogue should operationalize. The added value the AI Dialogue can uniquely bring is its universal UN membership legitimacy - giving it the convening authority to translate multi-stakeholder discussions into binding international commitments, which purely technical bodies like ITU or multi-stakeholder forums like IGF cannot do alone. The Dialogue should serve as the apex coordination mechanism that synthesizes these various threads into a coherent, action-oriented global AI governance architecture with clear accountability and implementation timelines.

The AI Dialogue should adopt a multi-track participation model that meaningfully integrates government, civil society, private sector, academia, and technical communities - not as separate silos but as interdependent voices in a structured deliberative process. For civil society and community organizations, the Dialogue should establish a dedicated civil society forum running in parallel to the intergovernmental sessions, with a formal channel for civil society inputs to be tabled and responded to in plenary. This mirrors best practice from UNFCCC COP processes. For private sector participants, particularly small and medium-sized tech enterprises from the Global South, the Dialogue should create an innovation showcase and policy dialogue mechanism - ensuring that African AI startups and platforms like Shecode.ai can share real-world evidence of both the opportunities and governance failures they encounter. For academia, the Dialogue should commission thematic research briefs from universities in the Global South to provide evidence-based inputs that balance the often Northern-centric research base that dominates AI governance discourse. Structurally, sessions should be hybrid-first - enabling remote participation for stakeholders from developing countries who cannot travel to Geneva, with simultaneous interpretation available in UN languages plus key regional languages such as Swahili, Arabic, and Hausa. Dedicated capacity-building sessions before the main plenary would help equip stakeholders from low-resource settings to engage substantively with technical governance questions, rather than arriving unprepared and therefore disempowered.

• The most critically underrepresented voices in global AI governance are: women and girls, particularly those in the Global South
• indigenous and rural communities whose languages, livelihoods, and cultural practices are being shaped by AI systems they had no role in designing
• young people aged 18-35 in Africa, Asia, and Latin America who are simultaneously AI's most active users and most underrepresented in governance spaces
• and persons with disabilities, who face acute AI-driven exclusion in digital service access. From my work with PAWA Initiative and Shecode.ai, I see daily how African women tech practitioners are excluded from global governance conversations not due to lack of expertise but due to structural barriers - visa restrictions that prevent travel to Geneva or New York, lack of funding to attend international forums, and online participation mechanisms that are poorly designed for low-bandwidth environments. To address this, the AI Dialogue should: establish a sponsored fellowship for civil society delegates from underrepresented regions and demographics to attend in person
• create a digital participation platform optimized for low-bandwidth environments
• partner with regional civil society networks across Africa, South and Southeast Asia, and Latin America to convene grassroots consultations whose outputs are formally submitted to the Dialogue
• and mandate gender parity and regional diversity requirements for speakers and panelists in all sessions. Inclusion must be structural, not cosmetic - diversity in the room means nothing if the governance outcomes continue to reflect only the priorities of those who have historically dominated the conversation.

To foster genuinely dynamic engagement, the AI Dialogue should move beyond the traditional plenary-and-panel format and experiment with several innovative models. First, thematic working groups with formal civil society co-chairs should be convened between sessions, producing substantive written recommendations rather than simply commenting on government-drafted texts. Second, a citizen and community testimony segment - modeled on human rights council mechanisms - should be embedded in each major session, where community representatives directly impacted by AI systems (women farmers using AI-based weather apps, gig workers managed by algorithmic platforms, students subject to AI proctoring tools) can give formal testimony that is entered into the official Dialogue record. Third, a real-time translation and accessibility protocol should be implemented, ensuring that participants from non-English-speaking regions can engage authentically in their preferred languages. Fourth, a pre-Dialogue grassroots consultation process - using mobile-first digital tools accessible even on low-data connections - should gather inputs from communities across the Global South, aggregated into regional position papers. Fifth, a youth-led side summit running in parallel to the main sessions would create a pipeline of next-generation AI governance voices and produce a Youth AI Governance Compact submitted formally to the plenary. These formats would transform the AI Dialogue from a diplomatic conference into a living, participatory governance process that reflects the full diversity of humanity's relationship with AI.

Several promising practices and platforms deserve global attention as models for effective AI governance. Rwanda's National AI Policy and its focus on using AI for agricultural productivity and healthcare delivery in a resource-constrained context demonstrates that developing nations can be AI governance innovators, not just adopters. Nigeria's National Information Technology Development Agency (NITDA) has developed an AI policy framework with a dedicated Ethics and Governance pillar - providing a template for African nations seeking to build regulatory capacity. At the civil society level, the Algorithmic Justice League's work on auditing AI systems for gender and racial bias provides a practical methodology for bias detection that should be adopted as a global standard within AI governance frameworks. The Mozilla Foundation's Trustworthy AI program and its community-based research model demonstrates how civil society can produce credible, technically rigorous AI governance inputs. At the platform level, Shecode.ai's work in Nigeria on equipping women with AI literacy and entrepreneurship skills represents a grassroots approach to closing the AI gender gap that could be scaled with international support. The EU's AI Act, while imperfect, offers a risk-tiered regulatory approach that developing nations could adapt to their local contexts. The key lesson from all effective AI governance practices is that they are contextually grounded, multi-stakeholder in design, and adaptive over time - rather than one-size-fits-all frameworks imposed without local validation. The AI Dialogue should document, disseminate, and build upon these practices, creating a living repository of AI governance innovations from across all regions.