China and US Diverge on Global AI Access: Competing Governance Models Crystallize
Key Takeaways
- ▸China proposed a global AI cooperation organization offering free/cheap open-weight models, explicitly positioning itself as an alternative to the US-led 'trusted partner' approach
- ▸The US and G7 are formalizing a coalition-based governance model with restricted access to advanced American AI, with Google DeepMind and Anthropic actively advocating for this framework
- ▸The US Commerce Department's decision to cut off foreign access to Anthropic's models exemplifies the restrictive approach China is now directly challenging
Summary
A fundamental divergence in global AI governance is crystallizing, with China proposing an open, internationally-accessible AI cooperation organization offering free or cheap models to developing nations, while the United States and G7 are formalizing a restricted "trusted partner" coalition with controlled access to advanced American AI systems. The timing is pointed: the US Commerce Department recently ordered Anthropic to shut down its Fable 5 and Mythos 5 models to all foreign users, while Chinese government officials explicitly criticized "closed, exclusive and monopolistic approaches" to AI development. The G7 summit in France saw Anthropic and Google DeepMind advocate for a US-led international AI coalition, even as China positions itself as the provider of accessible AI to the Global South.
The two approaches could not be starker. Chinese models like DeepSeek and Qwen are available as open-weight downloads to anyone with internet access, while US models like Anthropic's are increasingly subject to export controls and subscription-based access restrictions. China is leveraging existing multilateral structures (BRICS, Shanghai Cooperation Organisation) and new initiatives like "AI Capacity Building for All" to position itself as the accessible alternative. The US approach prioritizes alliance-building and rule-setting among wealthy democracies, leaving 6 billion people outside the G7 to choose between two fundamentally different systems: one that treats AI as an open platform, and one that treats it as a gated resource for trusted partners.
For developing economies and the majority of the world's population excluded from the G7's framework, this competition is consequential—it will determine not just which AI models they can access, but the governance principles and geopolitical alignments that shape global AI development for the coming decade.
- The Global South, excluded from G7 negotiations and unable to afford enterprise AI subscriptions, now faces a binary geopolitical choice about which AI system to adopt
