AI Trust and Safety in Emerging Technology Roundtable

Summary

The AI Trust and Safety in Emerging Technology Roundtable hosted in Abuja, Nigeria was convened to create a collaborative space for critical dialogue on the role of trust, safety, and accountability in the development and deployment of AI systems within African contexts. The event aimed to move beyond global narratives by interrogating what responsible AI truly means in environments shaped by unique social, economic, and regulatory realities.

Specifically, the roundtable sought to challenge existing assumptions, surface under-discussed risks such as bias, misinformation, and data privacy, and explore how these issues manifest across Africa’s rapidly evolving technology ecosystem. By bringing together a diverse group of stakeholders including start-up founders, policymakers, researchers, and civil society actors, the event fostered cross-sector exchange and shared learning.

Designed as a closed-door, peer-level roundtable, the format encouraged open, honest conversations and active participation. Attendees contributed insights based on their roles within the ecosystem, helping to identify gaps, tensions, and opportunities for more responsible AI development.

Ultimately, the purpose of the event was to co-create practical, actionable, and locally relevant AI safety principles, while strengthening collaboration among stakeholders working to build ethical and inclusive innovation ecosystems across Africa.

Highlights

One of the most encouraging outcomes of the AI Trust and Safety in Emerging Technology Roundtable was the collective shift from abstract concerns about AI risks to practical, actionable solutions tailored for African contexts.

Bringing together startup founders, policymakers, researchers, and civil society actors, the session created a rare space for honest, cross-sector dialogue. What stood out was not just the diversity of perspectives, but the willingness of participants to challenge assumptions and rethink responsibility in building safe AI systems.

A key highlight was the co-creation of practical AI safety principles grounded in African realities. Rather than adopting global frameworks wholesale, participants emphasized the need for approaches that are feasible for startups, equity-centered, and enforceable within existing regulatory constraints. This marked an important step toward developing locally relevant standards for responsible innovation.

Another major takeaway was the reframing of accountability. The discussion expanded beyond the traditional focus on builders to include investors and funders as critical actors in ensuring AI safety. This shift signals growing recognition that responsible AI is not just a technical challenge, but a systemic one that requires shared responsibility across the entire ecosystem.

Importantly, the event also showcased SafetyMeter, an AI-powered tool designed to help startups identify risks early and integrate safety-by-design into their products. This demonstrated that ethical considerations can be embedded into innovation processes without slowing progress.

Overall, the roundtable demonstrated that Africa is not just responding to global AI trends, but actively shaping its own approach to ethical and responsible AI. The conversations sparked, partnerships formed, and commitments made are a strong foundation for continued collaboration and action.

If sustained, these efforts have the potential to influence how AI is built, governed, and trusted across the continent.

Impact

Key Findings and Impact of the Project

The AI Trust and Safety Roundtable surfaced several critical findings on the current state of AI adoption, governance, and risk in African contexts. A key insight was the significant gap in AI literacy, where many users and organizations still perceive AI primarily as chatbots rather than complex systems influencing decision-making, automation, and data processing. This gap contributes to misuse, overreliance, and increased exposure to risks such as misinformation and unsafe data practices.

Another major finding was the lack of robust AI governance frameworks across African countries, with limited enforcement mechanisms and heavy reliance on global standards that are not always contextually relevant. Participants also highlighted emerging risks including “shadow AI” in organizations, data privacy vulnerabilities, bias due to underrepresentation of African data, and growing threats from misinformation and synthetic content.

A particularly important insight was the multi-stakeholder nature of responsibility in AI systems, with debates emphasizing that accountability must extend beyond builders to include investors, governments, and civil society. This reflects a shift toward more systemic approaches to AI safety and governance.

Impact Created and Potential Future Impact

• Social Impact: The project raised awareness of AI risks affecting vulnerable groups, including children and low-literacy populations, and emphasized the need for inclusive and equitable AI systems. This contributes to improved digital safety, trust, and user wellbeing.

• Economic Impact: By introducing tools like the SafetyMeter and promoting safety-by-design, the project supports startups in reducing long-term risks, avoiding reputational damage, and building more sustainable and investable products.

• Organizational Impact: The roundtable strengthened collaboration across sectors and initiated partnerships with developer communities, creating pathways for capacity building and improved decision-making in AI development processes.

• Policy and Sector Impact: The discussions highlighted urgent gaps in governance and contributed to shaping conversations around actionable AI policy frameworks tailored to African realities, with potential to influence future regulatory approaches and industry standards.

Looking ahead, the project has strong potential to drive long-term ecosystem change by embedding AI safety practices into product development, informing policy discussions, and fostering sustained multi-stakeholder collaboration.

Lessons learned

Several important lessons emerged from planning and facilitating the AI Trust and Safety Roundtable.

First, the value of a closed-door, in-person, peer-level format was particularly evident. Bringing participants together physically created a more engaging and focused environment, allowing for deeper conversations, trust-building, and more candid exchanges especially on sensitive topics such as accountability, data misuse, and organizational risks. This format proved more effective than traditional panels for generating honest, high-quality insights.

Second, the event highlighted the need for stronger foundational AI literacy across stakeholder groups. Even among professionals, there were varying levels of understanding about AI systems and their broader implications. This reinforces the importance of embedding awareness and capacity-building into future engagements.

A key operational and strategic lesson was the challenge of engaging government and public sector stakeholders. There is significant bureaucracy involved in securing their participation, particularly for independent or emerging organizations. Unlike events backed by large global technology companies such as Meta or Google, where there are often clear financial or institutional incentives, smaller convenings may struggle to attract public sector attendance. This highlights the need for more intentional engagement strategies and possibly formal partnerships to ensure government representation in future discussions.

Additionally, the session emphasized the importance of balancing technical and policy perspectives. While the diversity of participants enriched the conversation, it required careful facilitation to ensure clarity, alignment, and productive dialogue across disciplines.

Finally, there was a clear need to move from dialogue to sustained action. Participants expressed strong interest in follow-up engagements, including training, partnerships, and practical implementation of tools like SafetyMeter. This shows the importance of designing events with clear post-event pathways to maintain momentum and translate insights into impact.

Outcomes & outputs

The AI Trust and Safety Roundtable generated several key outputs and outcomes that will contribute to research, practice, and collaboration in AI governance and safety within African contexts:

• Comprehensive Event Report: A detailed report capturing key insights, risks, discussion themes, and strategic recommendations on AI trust, safety, and governance in Africa.

• Documented Key Findings and Risk Areas: Identification of critical challenges including AI literacy gaps, data privacy risks, misinformation, bias, “shadow AI” in organisations, and gaps in governance frameworks.

• SafetyMeter Demonstration and Feedback: Showcase of the SafetyMeter tool as a practical solution for integrating safety-by-design into AI products, along with participant feedback for further refinement and adoption.

• Multi-Stakeholder Dialogue and Knowledge Exchange: Facilitated cross-sector engagement between startups, policymakers, researchers, and civil society, strengthening shared understanding of AI risks and responsibilities.

• New Collaboration Pathways: Initiation of partnerships with developer communities and organizations interested in follow-up trainings, tool adoption, and joint research on AI safety (With AI Safety Nigeria )

• Content for Public Engagement: Development of blog posts, social media content, and communication materials to disseminate insights and raise awareness on AI trust and safety issues.

• Foundation for Future Programming: Establishment of the roundtable as a recurring engagement, with plans for future sessions, capacity-building workshops, and research collaborations.