YAIL Kajiado Hub

In my opinion, the success of the first Global Dialogue on AI Governance will depend on one key outcome: radical honesty about the real risks and trade-offs of AI. Right now, there is a global rush to innovate, and many governments and companies are more focused on staying competitive than on asking hard ethical questions. A successful dialogue should create a space where stakeholders, governments, tech companies, researchers, and civil society, can openly acknowledge current and emerging risks such as bias, misinformation, data exploitation, and loss of human control. These issues should not be softened or avoided for political or economic convenience. Instead, they must be addressed directly and transparently. Another important outcome would be a shared global understanding that responsible AI development is not a barrier to innovation, but a foundation for sustainable progress. If all stakeholders in the industry can agree that ethics and governance should move at the same speed as technological advancement, that would already be a major win.

• Safe, secure and trustworthy AI
• Open-source software, open data and open AI models
• AI capacity-building

First, safe, secure, and trustworthy AI is non-negotiable. We are building systems that can influence decisions at scale, and without strong safeguards, the risks can easily outweigh the benefits. From my perspective, safety should not be an afterthought, it should be part of the design process from day one. AI capacity-building is also critical, especially for regions that are often left behind in global tech conversations. If we don't intentionally invest in skills, education, and infrastructure, we risk creating a world where only a few countries shape AI while others are forced to adapt without influence. Transparency, accountability, and human oversight matter because AI systems are not neutral. They reflect the data and intentions behind them. If we don't make these systems understandable and hold people accountable for their outcomes, we create space for misuse and harm without consequences. Lastly, I included open-source software, open data, and open AI models because openness drives innovation and inclusion. However, this openness must be balanced with responsibility to prevent misuse.

One major gap is the speed of AI development versus the speed of governance. Right now, innovation is moving much faster than regulation, and that imbalance creates real risk. It's not just about having governance frameworks, but whether they can actually keep up with how fast AI is evolving. If this is not addressed, even the best policies will always be behind the problem.

In Kenya, the governance gaps in AI are already showing both serious challenges and real opportunities, especially from a software engineering and research perspective. One of the biggest challenges is the gap between adoption and regulation. AI is already being used in sectors like fintech, agriculture, and healthcare, but governance frameworks are still developing. This creates risks around data privacy, biased systems, and misuse of AI, especially when solutions are imported and not fully adapted to local realities. Without strong oversight, we may end up scaling systems that don't fully understand or represent Kenyan contexts. Capacity-building is another major issue. While Kenya has a growing tech ecosystem, there is still a shortage of advanced AI skills and research infrastructure. This limits our ability to build, audit, and govern AI systems locally. As a result, we often depend on external technologies, which increases the risk of misalignment with our social and economic needs.

One example is the EU AI Act. What stands out is its risk-based approach, AI systems are not treated the same, but regulated based on their potential impact. This is a practical way to balance innovation and safety instead of slowing everything down. It's not perfect, but it shows that governance can be structured and enforceable. On the African side, the African Union has been pushing forward AI strategies that emphasize local context, data sovereignty, and capacity-building. This is important because governance should not just be imported; it needs to reflect African realities. From a more practical and technical angle, open-source platforms like Hugging Face are helping improve transparency and collaboration. When models and datasets are open, researchers can audit, improve, and adapt them more easily. But again, this has to be balanced with safeguards to prevent misuse.