China Rejects US AI Pacing Push as No. 2 Race Intensifies

For AI and machine learning practitioners, the US-China clash over frontier pacing is not just geopolitics. It will determine whether Chinese labs keep closing the model-capability gap under export constraints, whether Western labs coordinate release schedules with governments, and whether safety research can remain a neutral technical field rather than a lever of strategic competition.

The intensifying US-China artificial intelligence contest entered a sharper phase after Anthropic chief executive Dario Amodei publicly floated a "worldwide pacing of the frontier" and argued that any such slowdown would require cooperation with China. His proposal, framed around preventing catastrophic loss of control of the most advanced AI systems, lands as Silicon Valley's mood has turned somber, with insider after insider warning that frontier model capabilities are accelerating beyond established safety frameworks. But Beijing's response has been unequivocally negative. According to Lian Jye Su of technology research group Omdia, China is "definitely not receptive" to a slowdown on US firms' terms, because the government sees AI as a "core sovereignty capability" and interprets the calls as a bid to lock in US advantage. The result is not a simple policy disagreement: it is a collision between two fundamentally different theories of how to balance AI pace and safety.

“

The United States frames its actions as responsible risk management; China frames the same actions as strategic containment.

The Chinese position is rooted in concrete economic and strategic stakes. For roughly two years, Chinese businesses across the economy have been pushed to integrate AI before being made obsolete by it. The education entrepreneur Boris But described his own "penny-drop moment" when wealthy clients began arriving with opinions gathered not from rival education consultancies, but from chatbots. This bottom-up adoption is being actively encouraged by the state as a growth tool, and it is paired with an official view that AI is tied directly to national sovereignty. US export controls on advanced chips and machinery already amount to what some see as a forced unilateral pacing of Chinese development. In that context, Beijing reads Amodei's proposal not as a neutral safety initiative, but as an effort to freeze the current hierarchy in which US frontier models remain ahead of their Chinese counterparts.

Analysts generally agree that frontier US models are still some way ahead, but the gap is narrowing. That narrowing is precisely why the debate is becoming more urgent. From Beijing's perspective, agreeing to a joint slowdown now would trade away its hard-won progress. Amodei made the competitive logic explicit when he argued that pacing should keep "democracies' AI lead over autocracies as large as possible", and he advocated continued export controls on advanced chips and machinery. For Chinese policymakers, this confirms that safety language is inseparably linked to US industrial and strategic policy. China's foreign ministry responded directly to Amodei's piece, rejecting the framing that Chinese development is uniquely dangerous and defending its own balance between safety and pace.