Who Sets the Rules for AI?

Multilateral coordination remains fragmented

The UN dialogue has produced a coordination roadmap and proposals for shared safety baselines, interoperability, and cross-border testing. Yet national approaches remain divergent, and competing geopolitical priorities and limited support for binding rules keep coordination largely voluntary.

Drawn from 5 articles

Implementation and assurance lag deployment

Governance measures and safeguards remain unevenly implemented: sector-specific strategies, liability standards, and regulatory guidance are incomplete or absent in parts of the corpus. Existing frameworks therefore leave substantial reliance on voluntary measures and variable national execution.

Drawn from 3 articles

Capacity and control remain concentrated

AI development, evaluation infrastructure, and governance influence remain concentrated among firms and countries with greater resources, limiting many countries’ ability to assess systems or shape rules. Funding and capacity-building proposals address participation gaps, but the evidence also raises concerns that they may leave dependence on dominant infrastructure intact.

Drawn from 4 articles