Page 76 - The Annual AI Governance Report 2025 Steering the Future of AI

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

Chapter 3: Regional Perspectives and Case Studies

The Dialogue featured several concrete examples of how countries and regions are beginning
to address the governance challenge. These case studies illustrate both diversity of approach Regional
and potential for cross-learning. Chapter 3:

3.1 European Union

Europe’s approach to AI governance is characterized by a risk-based approach, distinguishing
between unacceptable, high, and low-risk systems. The European Union’s AI Act was repeatedly
referenced as the first comprehensive, binding framework for AI. By adopting a risk-based
approach, the Act seeks to balance innovation with safety and trust, providing a model that
others may adapt.

In Europe, trust is seen as the foundation of AI adoption. The EU sees hard law as a necessary
way to build that trust. Transparency measures such as registries – Estonia’s AI registry was
cited as a pioneering national initiative, offering citizens and policymakers visibility into how AI
is used in public administration –, model documentation, a mechanism for quick updates, and
a just published (10 July 2025) code of practice were highlighted as mechanisms for building
this trust. At the same time, Europe has invested in innovation sandboxes, allowing developers
and regulators to collaborate in testing new systems before large-scale deployment.

The purpose of the General-Purpose AI Code of Practice, a voluntary guide developed by
independent experts through a multi-stakeholder process, is to help companies meet the
European AI Act's legal requirements concerning the safety, transparency, and copyright of
general-purpose AI models. Both European Member States and the European Commission
are now evaluating the code.

67