Page 86 - The Annual AI Governance Report 2025 Steering the Future of AI

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

4.2 Co-chairs’ Ten Pillars for AI Governance

1. From Principles to Practice AI governance should move beyond high-level declarations
to practical implementation that enables sustainable innovation and long-term impact. for 2026
Agile and inclusive frameworks, adaptable oversight mechanisms, technical standards Chapter 4: A vision
and tools are essential to guide AI development and deployment in ways that are socially,
economically, and environmentally responsible.

2. A Multistakeholder Imperative Governance that affects all should be shaped by all.
Governments, civil society, academia, the private sector, technical experts, and inter-
national organizations should co-create policy. All countries need a seat at the table,
supported by capacity building, so that AI can benefit everyone, everywhere.

3. Transparency as a Cornerstone of Trust Understanding how AI systems are built, eval-
uated, and used is important. Transparency in model behavior, data practices, and
decision-making processes strengthens accountability, builds public confidence, and
unleashes responsible innovation.
4. Bridging Inclusion AI governance should reflect diverse perspectives. Bridging the digital
divide through inclusion goes beyond access—it means enabling meaningful participation
in shaping the technologies and rules that affect people’s lives.

5. Capacity for All, Not Just a Few Closing global gaps in AI readiness is critical. Capaci-
ty-building initiatives—spanning policy advice, skills training, institutional strengthening,
and financial support—should empower communities worldwide to govern AI effectively
and to innovate in key sectors such as health, education, and agriculture.

6. Environmental Sustainability and AI Infrastructure Sustainable AI development should
address its environmental footprint—energy, water, and resource demands. Governance
frameworks should integrate energy and environmental policies, promote efficient data
centers and renewable power, and ensure AI projects can scale without overstraining
local infrastructure.
7. Sectoral Focus and Broad Collaboration The value of AI is realized through its appli-
cations in health, education, agriculture, humanitarian assistance including in disaster
management, and many other critical areas. Governance should involve respective
communities, adopt a cross-government and cross-society approach, and leverage inter-
national frameworks so that AI can deliver targeted benefits and address sector-specific
challenges.

8. Standards and Safety Tools Technical standards, benchmarks, and audit protocols
are foundational for safe, interoperable, and agile AI governance. Developed through
international multistakeholder processes, these tools should be evidence-based and
adaptable to rapid technological evolution.
9. Governance of Compute and Models As AI models scale in capability, governance of
compute resources and large foundation models becomes more important. Access to
compute infrastructure, robust risk assessments, and accountability frameworks ensure
that powerful AI systems serve the public interest.

10. Policy Interoperability and Agile Governance Coherent and interoperable policy frame-
works prevent fragmentation while providing clear policy direction. Agile governance
– integrating adaptable rules and inclusively developed technical standards – enables
flexible adaptation to technological advances.

77