Research Summary: Mapping General-Purpose AI Governance in Twenty AI Middle-Power Jurisdictions

Research indicates that 'AI middle-powers' – advanced economies without frontier General-Purpose AI (GPAI) developers – are independently establishing governance frameworks for GPAI. This analysis maps provisions across twenty jurisdictions, including the European Union, focusing on systemic risk assessment, evaluation, prohibitions with monitoring, and serious incident reporting to establish accountability for the model layer. The study notes that while GPAI development is concentrated in two major global powers, the risks associated with GPAI are global, prompting these middle-powers to develop their own regulatory responses.

Why it matters

The proliferation of national-level GPAI governance frameworks by 'AI middle-powers' signals a complex and fragmented regulatory landscape emerging globally, despite the concentrated development of GPAI itself. This decentralised approach to regulation will significantly impact international technology standards, market access, and the operational compliance burden for entities developing, deploying, or using GPAI models across different regions.

Key insights

• General-Purpose AI (GPAI) model development is primarily concentrated in two jurisdictions: the United States and China.
• The global risks associated with GPAI necessitate governance efforts from a broader range of nations.
• Twenty 'AI middle-power' jurisdictions, defined as advanced economies without frontier GPAI developers, are actively creating their own GPAI governance rules.
• The governance frameworks being developed by these middle-powers focus on four key areas: systemic risk assessment, evaluation and verification, prohibitions with monitoring and detection, and serious incident reporting.
• The analysis maps individual provisions within these governance areas to understand the accountability chain for the model layer of GPAI.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2608.19278