Would China agree to slow AI development? That’s the wrong question

The article suggests that focusing on whether China would agree to slow AI development might be the "wrong question." Instead, it proposes examining who constitutes "China" in such discussions. The AI safety community within China, for instance, is expected to share concerns about frontier risks, as indicated by the AI Safety Governance Framework 3.0 recently published by the National Technical Committee 260 on Cybersecurity of the Standardisation Administration of China.

This distinction is important because it highlights that "China" is not a monolithic entity with a single view on AI development. Different stakeholders, such as technical experts and government bodies, may hold varying perspectives and priorities regarding AI safety and the pace of development. Understanding these internal dynamics is crucial before predicting any national stance on international agreements.

The article implies that for meaningful US-China action on frontier AI, two elements are necessary: a secure environment for technical experts from both sides to collaborate, and pressure on leading companies to implement responsible practices. This suggests a shift from high-level political negotiations to more granular, expert-driven engagement and corporate accountability.