Sandboxes are popping up all over the tech space. In just a few years, they’ve moved from niche experiments to central tools in major policy initiatives, from the EU AI act to the United State’s newly released AI Action Plan. At the Datasphere Initiative, we’ve spent years exploring how sandboxes are designed, run, and evaluated. There’s one question we hear again and again: do they actually make a difference?
Evidence from a UK sandbox experience, for example, shows that firms entering the sandbox saw a 15% increase in capital raised and a 50% higher probability of raising capital, with a significant positive effect on survival rates and patenting.1 But the reality is, impact can be measured in different ways and depends on the goals of the sandbox.
Sandboxes real value relies on the actors who set the sandbox up and the participants who ultimately engage in them.
With that in mind, this blog takes a look at some examples of the impact sandboxes have had: from supporting emerging businesses scale, providing regulatory clarity, shaping regulatory direction or even supporting access to crucial services such as health-care and electricity.
Unsure what a sandbox is? See our 5-min intro to Sandboxes
Supporting emerging businesses to scale
When many hear the word sandbox, they think of the financial sector. That’s likely due to the UK’s FCO leadership in sandboxes since 2016. Reading one of their many public reports on their sandbox outcomes one can derive a range of examples of impact (as well as lessons learnt) and how these sandboxes are helping innovative business models in the financial sector.
For example, one report shows that access to the regulatory expertise the sandbox offered had reduced the time and cost of getting innovative ideas to market. An area of consideration for investors is the regulatory certainty of the firms they are considering investing in.
Feedback from one firm stated that continuous dialogue with the FCA during the sandbox process enabled them to develop more rigorous policies and processes than they otherwise would have.
Providing regulatory clarity
When Norway started to implement the EU’s GDPR some businesses were worried about complying with these new rules and what this would mean for innovation.
Norway’s data protection authority set up the Regulatory Sandbox in 2020 under the National Strategy for Artificial Intelligence.
This sandbox has led to businesses ensuring that solutions not only complied with data protection laws but also set a precedent for privacy-friendly innovations in the security industry, exemplifying how sandbox participation can lead to the successful integration of privacy principles in technological advancements.
Shaping regulatory guidance
As a global technology and innovation hub, Singapore wanted to get ahead with adopting privacy enhancing technologies. But many companies weren’t convinced and felt their use could put their compliance with data protection laws at risk.
The IMDA Data Privacy / Privacy Enhancing Technologies (PETs) sandbox aimed to help companies explore together with the regulators, whether PETs could enable secure, privacy-respecting data collaboration in domains where traditional data sharing is often constrained by regulatory or ethical barriers.
Several participants in the Singapore PET Sandbox developed proof-of-concept solutions that applied advanced privacy-preserving technologies to real-world data challenges across different sectors.
Beyond technical findings, the sandbox revealed the importance of early alignment between legal teams, data stewards, and technology developers. The sandbox served as a trust-building platform, allowing participants to articulate shared risks, expectations, and accountability structures.
Supporting initiatives to provide access to crucial services such as health-case
In 2023, the Ministry of Health launched a regulatory sandbox to support the development of the digital health innovation ecosystem and telemedicine was included.
Much like the regulatory sandbox model applied to Fintech, the sandbox was a testing mechanism for digital health innovation led by the Ministry of Health in collaboration with various experts, and included both the public and private sector.
The regulatory sandbox assessed the reliability of business processes and models.
The sandbox also encouraged service providers and regulators to jointly explore innovative business models and analyze risks to society while allowing the development of regulations that actively support the implementation of these new technologies in the health sector.
The test process generated learnings that resulted in solid recommendations for the development of evidence-based regulations in the health sector and reportedly supported hospitals to strengthen consumer protection and patient safety, including personal data of service users.
These examples just skim the surface of the case studies and research the Datasphere Initiative is developing on sandboxes and their impact in supporting innovation, boosting AI adoption and facilitating evidence based policymaking.
Another crucial element of regulatory sandboxes that are often left out of the conversation on impact, is their their preventive effect. Beyond enabling innovation, sandboxes actively help minimize risks. Once a sandbox is in place, not only do participants benefit, but others in similar positions are nudged to make more responsible choices. This becomes particularly relevant when dealing with emerging technologies, where formal regulations and clear boundaries are not yet in place.
A challenge often faced is how to report on sandboxes and share learnings that are sometimes sensitive and often experimental.
Through the Global Sandboxes Forum and our trainings on how to design, participate and implement sandboxes, we hope to build this field of knowledge: showcasing what worked and providing a safe space to share what didn’t.
1 Cornelli, G., Doerr, S., Gambacorta, L., & Merrouche, O. (2021). Regulatory Sandboxes and Fintech Funding: Evidence from the UK. Regulation of Financial Institutions eJournal. https://doi.org/10.2139/ssrn.3727816.
