Devising Regulatory Sandboxes and Responsible Practices for Designing AI-based Services in the Finnish Public Sector

In this paper, we examine the opportunities and challenges of devising mechanisms for responsible AI practices in the public sector. We discuss how innovations in public sector AI must comply with suitable ethical and regulatory frameworks, while fostering experimentation and participation of diverse actors throughout the AI lifecycle of designing, deploying, and assessing such services. We examine AI-based innovations and ethical practices in the context of prototype systems and end-user services provid-ed by two Finnish public sector organizations. The paper proposes regula-tory AI sandboxes engaging multiple stakeholders and integrative frame-works (MLOps and RegOps), to facilitate experimentation, co-learning, and deployment of future public sector AI services with ethical practices and regulatory compliance. Upcoming regulations such as the AI Act and the AI Liability Directive make the adoption of such sandboxes beneficial as a space for limited experimentation, that goes beyond ethical considerations to specific legal compliance and allocation of liability. In the future, this framework must be validated through pilot experimentation with pub-lic/private stakeholders and regulators in different areas of high-risk AI ser-vices.