Page 57 - The Annual AI Governance Report 2025 Steering the Future of AI

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

1.4 Power Concentration

Another recurring theme was the concentration of power. A small number of corporations
and countries control the majority of compute infrastructure, talent, and datasets required
for frontier AI development. This concentration not only raises concerns about monopolistic
practices but also about accountability: decisions made in a handful of boardrooms could
shape the trajectory of AI globally. Governments often lack the power people attribute to them,
as powerful corporations increasingly influence politics, said Greece’s former Prime Minister
George Papandreou. Despite AI's potential to decentralize, it is currently leading to further
concentration of economic and wealth power in the hands of tech giants and oligarchs, who
often argue that regulation would stifle innovation.

Some panelists worried that without intervention, the benefits of AI will accrue to those already in
positions of power, while risks are externalized to the broader public. The imbalance of influence
could hinder competition, limit innovation in smaller markets, and entrench digital colonialism.

At the same time, some saw opportunities in shared infrastructure initiatives. Proposals such as
international compute facilities, collaborative research hubs, and open-source model registries
could democratize access. By reducing dependency on a small set of actors, such initiatives
could help balance the concentration of power with distributed innovation.

Quotes:

• “[B]ig corporations, huge powers [are] influencing politics and that of course
puts governments in a very difficult situation. [Governments] do not really have
power that people think [they] do”. (George Papandreou, former Prime Minister
of Greece, General Rapporteur for Democracy, PACE, Council of Europe)
• “Some of the unchecked risk taking by a few elites could actually lead to systemic
risk on a global level, including for the Global South and unprotected communities
worldwide.” (Brian Tse, CEO, Concordia AI)

Dive deeper in the Whitepaper “Themes and Trends in AI Governance”:
• 5.1 Global Compute Distribution

1.5 Complexity

AI presents not one, but many interacting problems (alignment risk, adversarial abuse,
accidents, ecosystem issues like labor and over-reliance), making a single, comprehensive
governance framework difficult to create. Legal and governance solutions require the technical
ability to monitor AI capabilities, which is currently severely limited given the complexity and
continuous evolution of AI systems.

Reducing complex issues to vague terms like "trust" or "safety" is not seen as helpful. Terms like
robustness, fairness, privacy, and traceability each demand distinct technical methods and
oversight, said Anja Kaspersen (Director for Global Markets Development, Frontier Issues and

48