Page 12 - The Annual AI Governance Report 2025 Steering the Future of AI

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

At the AI Action Summit in Paris (2025), Countries from the Global Majority including India,
along regional bodies such as the African Union, played an active role in shaping the Summit’s
outcomes. Still, only a small number of countries dominate agenda-setting, while any countries
– over 100 – have no meaningful voice in AI governance. Part I – White Paper

Risk assessment has become a focus of AI governance, with growing efforts to institutionalize
evaluation practices and build shared safety infrastructure. AI Safety Institutes and equivalent
bodies are tasked with model testing, red teaming, and developing national safety protocols.
Alongside these institutions, multilateral bodies have begun coordinating on baseline risk
assessment methodologies, including proposals to standardize thresholds for system
classification, misuse potential, and post-deployment monitoring. However, approaches to
risk classification remain uneven, and few tools currently address real-world incidents at scale.
Governance of open-weight models remains contested, with proposals ranging from licensing
regimes to tiered release based on misuse potential.

Verification – the ability to confirm or validate another party’s actions or claims – is playing an
increasingly important role in AI governance by enabling trust, accountability, and enforcement.
For frontier AI, this includes validating training details, safety measures, system behavior, usage
during inference, and compute resources. By improving transparency, verification helps build
confidence and supports the development of shared global norms – much as it has in arms
control and climate agreements – turning voluntary principles into credible, coordinated action.

Introduction

The artificial intelligence landscape has experienced rapid technological advancement
throughout 2024 and early 2025. These developments include advanced capabilities in
autonomous agents, multimodel AI, robotics and expanded international coordination
mechanisms, significant economic deployment across sectors, and evolving governance
frameworks.

This white paper analyzes seven key themes emerging from recent AI developments: autonomous
agent deployment, verification systems, socioeconomic transformation, international
coordination, technical standards, infrastructure requirements, and risk management. The
analysis provides stakeholders with evidence-based insights into current AI trajectories and
their implications for policy and implementation decisions.

3