Output / Publikation
AI Sandboxes: Global Insights for Regulatory Learning and Adaptive Governance
This policy brief offers international insights into the design and function of AI sandboxes across Europe, Asia, Africa, and Latin America. It provides typologies, institutional design recommendations, and actionable guidance for policymakers seeking to develop sandboxes as flexible, rights-based, and context-aware governance tools.
As AI increasingly shapes decision-making in our daily lives, traditional governance frameworks struggle to keep pace. How can we ensure regulation keeps up with technological change?
In our latest policy brief “AI Sandboxes: Global Insights for Regulatory Learning and Adaptive Governance”, we explore diverse global practices and propose practical recommendations on how AI sandboxes—controlled, real-world environments for testing AI technologies, policies, and governance models—can support adaptive, inclusive, and anticipatory governance.
Key takeaways include:
- Four sandbox types (Operational, Regulatory, Hybrid, Policy), each suited for different governance contexts
- Practical case studies from Rwanda, Brazil, Singapore, and the EU
- Strategic recommendations for effective design, transparency, and international collaboration
- Insights into how sandboxes balance innovation with accountability, reducing risk before formal implementation
Read and download the full policy brief here:
Authored by Armando Guio, Markus Siewert, and Urs Gasser, this brief is a collaborative effort between the TUM Think Tank and the Global Network of Internet and Society Centers (NoC). Its insights draw on recent discussions with regulators, scholars, and practitioners from Europe, Latin America, Africa, and Asia at the workshop “AI Sandboxes: Regulatory Learning in Action” held at the UNESCO Global Forum on the Ethics of Artificial Intelligence in Bangkok (June 24, 2025). For a summary, see the report from the NoC here.
