AI governance splits into competing blocs

Global AI governance is becoming harder to achieve as AI systems, supply chains and risks cross borders while major powers compete for advantage. Development depends on globalised chips, GPUs, data centres and firms, and threats such as cyber-attacks, job displacement, pollution and water consumption are not confined to national jurisdictions. The race dynamic encourages states to delay or weaken rules to avoid falling behind.

The U.S. relies on a decentralised, sector-specific model that favours private-sector initiative and resists broad regulation, while pushing exports to allies and countering China in semiconductor controls and international bodies. China combines state-led domestic development with global proposals built around sovereignty, inclusiveness and sustainability, and has organised a UN coalition of 80 countries from the Global South. The EU, despite lagging in technical development, uses market access and the AI Act to advance transparency, rights and consumer protection.

Cooperation remains weak because each actor promotes incompatible governance models and treats AI leadership as a strategic prize. Past technology dilemmas, including ozone and nuclear governance, suggest agreements can emerge when fragmentation becomes too costly. For the EU, a more strategic approach to open AI could help reduce dependence on American platforms, compete with China’s open-source push and set clearer global terminology, while avoiding disclosure rules that burden European challengers.