Shaping the AI Sandbox Ecosystem for the Intelligent Age 2025

Appendix Institutional models for AI sandbox governance BOX A review of AI sandboxes established globally highlights the diverse objectives they serve. These objectives are often shaped by the mandate and priorities of the organization leading the sandbox. Building a comprehensive AI ecosystem requires the coordinated efforts of multiple sandboxes, each led by a different type of institution – government bodies, industry regulators, industry bodies, academia or private organizations – addressing distinct aspects of the broader challenge. The table below outlines the main categories of sandbox-leading organizations and their typical objectives. Leading stakeholder Typical objective Organization name Sandbox name Sandbox components Key purposeTypical sandbox objective acr oss stakeholders Global AI sandbox examples Put Cor nell at the for efront of AI innovation by providing faculty and staf f with access to AI tools Access to advanced GenAI models and security risks mitigation User -friendly platform for Stanfor d faculty , staff and students to safely try AI models Empower financial institutions to launch AI pr oducts rapidly and secur ely Accelerate AI r esear ch and development for gover nment applicationsDanish Data Pr otection Agency , DenmarkSupport the development of a robust AI ecosystem by enabling innovation thr ough access to compute and mentorship, accelerating adoption thr ough capability- building ef forts and of fering regulatory support Addr ess r egulatory challenges, support risk mitigation and enable compliant AI adoption in regulated sectors Enable sector -specific innovation and r esponsible use of AI Promote experimentation, foundational r esear ch and skill development thr ough secur e access to AI infrastructur e Accelerate AI pr oduct development, testing and scale-up thr ough secur e infrastructur e and industry collaborationGover nment Industry r egulator Industry body Academia Private/non- gover nmentMinistry of the Economy and Innovation of the Republic of Lithuania, Lithuania Infocomm Media Development Authority , Singapor e Malaysian Resear ch Accelerator for Technology and Innovation, Malaysia Ministry of Communication and Information T echnology , State of Qatar Medicines and Healthcar e Products Regulatory Agency , United Kingdom Financial Conduct Authority , United Kingdom Minnesota State Bar Association, USA National eLear ning Center , Saudi Arabia Harvar d University , USA Princeton University , USA Cornell University , USA Charles University , USA Stanfor d University , USA NayaOne, United Kingdom MITRE Corporation, USASupport AI innovation by pr oviding guidance on regulatory r equir ements Space for businesses to safely test innovations and ensur e regulatory compliance Support safe and trustworthy AI development and adoption Boost the adoption of AI thr oughout Malaysian organizations Flexible envir onment for developers, innovators and businesses to experiment with and develop AI-driven solutions Identify r egulatory challenges to AI as a medical device (AIaMD) and work collaboratively to understand and potentially mitigate risks Facilitate the development and launch of cutting-edge solutions within the financial services industry Contr olled envir onment for organizations to use LLMs to help impr ove access to justice Accelerate innovation and adoption of AI solutions in lear ning Secur e envir onment to experiment with GenAI Secur e envir onment for r esear chers with access to multiple LLMsRegulatory evaluation, mentorship Datasets, IDE, mentorship, market access IDE, LLMs, r egulatory evaluation Datasets, mentorship, funding LLMs, data analysis and visualization, code execution, IDE Compute, LLMs Compute, LLMs, data analysis and visualization Compute, LLMs, IDE LLMs, IDE IDE, datasets, testing and validation Compute, mentorshipRegulation guidance and expertiseRegulatory Sandbox for AI AI Sandbox GenAI Evaluation Sandbox AI Sandbox AI Sandbox AI Airlock Digital Sandbox AI Sandbox AI Sandbox in Digital Lear ning Harvar d AI Sandbox Princeton AI Sandbox Cornell SandboxAI Cuni AI Sandbox Stanfor d AI Playgr ound NayaOne AI Sandbox MITRE’ s Federal AI SandboxRegulatory assessment, compliance certification Testing and validation, regulatory evaluation Compute, datasets, testing and validation, capability-building, mentorship Integrated development envir onment (IDE), testing and validation, mentorship, capability-building Shaping the AI Sandbox Ecosystem for the Intelligent Age 24