On April 2, 2025, the Datasphere Initiative kicked off its AI Sandboxes Co-Creation Lab series in Kigali, Rwanda, ahead of the inaugural Global AI Summit on Africa. Held at Westerwelle Startup Haus Kigali, the session convened 17 participants from government, private sector, academia, development organisations and civil society to explore how sandboxes can serve as agile governance tools for advancing responsible AI across the continent. The lab was framed around the growing urgency for agile and adaptive governance frameworks to match the speed and complexity of AI and data-driven technologies. Download the Co-lab Report here.
Drawing from the Datasphere Initiative’s latest reports, Africa Sandboxes Outlook and Sandboxes for AI: Tools for a New Frontier, the workshop contextualized the evolving sandbox landscape in Africa and globally. The reports served as springboards for discussion, offering real-world examples of how sandboxes can support experimentation, enhance regulatory capacity, and serve as mechanisms for rights-respecting innovation.
Where Sandboxes Matter Most: Tackling Africa’s AI Challenges
Conversations at the Lab highlighted critical hurdles in Africa’s AI journey, most notably, the continent’s limited control over its own data and technology choices. Participants raised concerns over “digital colonialism,” where African data is extracted and processed elsewhere, with little benefit or oversight at home. This undermines trust and Africa’s ability to chart its digital future. Inconsistent regulation across countries also slows innovation and cross-border collaboration. Too often, compliance takes precedence over experimentation, missing the chance to test, learn, and co-create solutions.
In response, participants identified three critical areas where AI sandboxes could drive significant positive change across the continent.
- Data Governance and Digital Sovereignty: sandboxes can help African countries explore better ways to collect, manage, and govern their data. They allow for piloting new rules, frameworks, and standards, especially before laws are set in stone. Similarly, they enable testing of interoperable data-sharing models, which can be a smart alternative to strict data localization rules that limit innovation.
- Infrastructure and Contextual Solutions: with many regions facing infrastructure challenges, sandboxes create safe spaces to test tech like edge AI or other low-resource technologies that fit local realities. They also help build sustainable business models that go beyond one-off pilots, especially in sectors like health, agriculture, or education where AI can amplify the impact of digital public infrastructure.
- Skills and Capacity Development: sandboxes serve as practical learning environments by providing access to real datasets for research, experimentation and technical training. They help build technical capacity among African youth and practitioners while equipping regulators with the knowledge needed to evaluate and govern emerging AI systems.
Simulating Responsible Sandbox Models: From Concept to Practice
During the Lab, participants had the chance to bring sandbox concepts to life (regulatory, operational, or a hybrid), by tackling challenges previously identified. One team proposed a hybrid sandbox to address health data fragmentation and rural data access. Another explored a regulatory sandbox to improve institutional (particularly public institutions) understanding of emerging technologies including AI oversight.
Through interactive group work, participants defined clear goals, use cases, success indicators, and key players essential for implementing effective sandboxes. There was a shared understanding that sandboxes can reduce regulatory bottlenecks and fast-track responsible innovation by offering a structured, transparent space for testing ideas safely and collaboratively.
Additionally, responsibility emerged as a cornerstone of sandbox design. Participants co-created guiding principles to ensure ethical and inclusive practices, including:
- Demonstrating tangible benefits for all stakeholders.
- Designing frameworks grounded in research and academic partnership.
- Promoting open-source innovation while protecting intellectual property.
- Ensuring participation of communities most affected by AI applications.
- Addressing data risks through strong privacy and governance protocols.
- Use flexible responsibility frameworks that reflect local cultures and values while allowing consistent assessments across sectors and use cases.
Conclusion
The Kigali Lab represented a pivotal step toward a broader, Africa-led framework for agile and inclusive AI governance. With momentum building for future Labs and coaching programs, the Datasphere Initiative remains committed to supporting African policymakers and communities in transforming innovation into impact.
