Page 80 - The Annual AI Governance Report 2025 Steering the Future of AI

The Annual AI Governance Report 2025: Steering the Future of AI

3. Deploy AI Across Public Services

• Integrate AI into multiple sectors (education, healthcare, justice, transport, environment,
culture) – today, Estonia has about 200 AI applications across government institutions.
• Use AI for practical citizen services (e.g., virtual assistants, autonomous museum buses, Regional Chapter 3:
flood detection, AI learning tools).
• Scale beyond pilots — embed AI widely in government operations.
4. Adopt a Human-Centric Approach

• Design digital and AI services for citizen benefit first, not just government efficiency.
• Ensure citizens can see and control how their data is used.
• Maintain transparency tools (e.g., Algorithm Registry).

5. Balance Innovation with Responsibility

• Support risk-based regulation (e.g., EU AI Act model).
• Provide AI sandboxes, training, and competence centers to help stakeholders comply
and innovate.
• Safeguard digital rights and online freedoms alongside AI adoption.
6. Prioritize Education and AI Literacy

• Integrate AI learning tools into schools and curricula.
• Train both students and teachers in responsible AI use.
• Build societal resilience by fostering AI literacy early.
7. Commit to International Cooperation

• Share expertise and support other countries in building AI-ready societies.
• Participate in global AI governance initiatives (e.g., Global Digital Compact, Freedom
Online Coalition).
• Promote inclusive AI ecosystems to reduce global digital divides.

Quote:
• “The heart of Estonia’s previous digital transformation was the understanding that
we should not simply digitalise existing services but rather rethink and rebuild the
whole system.” (H.E. Mr. Alar Karis, President of the Republic of Estonia)

3.6 Switzerland

Switzerland, which sees itself as a leader in international AI cooperation and a bridge builder for
global collaboration, introduced two key Swiss initiatives designed to democratize access to AI:
• The Swiss-made Large Language Model: The model will be released under an open
license, with its source code, weights and training data all publicly available. A core
feature is its multilingual design – the model has been trained on datasets from over
1000 languages. The model is a flagship of the Swiss National AI Institute (SNAI), a
collaboration between researchers at the Swiss universities ETH and EPFL. The goal is
to create a legally compliant, accessible model that can serve as a foundation for various
applications in areas like healthcare, sustainability, and science. The model is being

71