Page 56 - The Annual AI Governance Report 2025 Steering the Future of AI

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

A good example of this approach is Switzerland's investment in the Apertus AI model – see
Chapter 3.6.

By developing Apertus, Switzerland has not achieved full AI sovereignty, but it has taken a
meaningful step toward it. This targeted investment has reduced its dependence on foreign Context Chapter 1: Global
models, given it greater control over specific use cases, and built valuable local expertise. The
goal of the project was not just to create a model but to fully understand how AI works, setting
a new standard for transparency and reproducibility in the field. This pragmatic approach shows
that nations can strengthen their position in the global AI ecosystem by focusing on achievable
goals rather than pursuing an impossible dream.

Figure 8: (from left) Robert Trager, Co-Director, Oxford Martin AI Governance Initiative,
University of Oxford; Lan Xue, Distinguished Professor and Dean, Schwarzman College,
Tsinghua University; Dawn Song, Professor of Computer Science at UC Berkeley and
Director of Berkeley RDI (Berkeley center for responsible decentralized intelligence;
George Papandreou, former Prime Minister of Greece, General Rapporteur for
Democracy, PACE, Council of Europe, Greece; Artemis Seaford, Head of AI Safety,
ElevenLabs

Quote:

• “There's a good chance we'll end up with AI systems that are superior to humans
across the board, not just comparable to humans. And what does that mean?”
(Yoshua Bengio, Full Professor at Université de Montréal, Co-President and
Scientific Director of LawZero and Founder and Scientific Advisor)

47