China rejects calls for ‘pacing’ on AI development, fearing it would entrench US tech lead

Calls for a voluntary slowdown in artificial intelligence development have been met with skepticism from Chinese researchers and state media, who believe such a proposal is an effort to bolster America's technological lead rather than a genuine initiative for responsibility. The debate escalated on Monday when OpenAI CEO Sam Altman and other Silicon Valley leaders urged industry-wide "pacing," warning that the rapid advancement of AI should not override safety considerations.

Altman argued that the world needs confidence in American companies' responsible development of increasingly capable AI systems, demanding that every frontier lab contribute to this responsibility. Similar calls came from Anthropic CEO Dario Amodei and xAI founder Elon Musk, who advocated for the global slowing down of frontier AI development, though they acknowledged the difficulty in achieving this.

However, Chinese researchers and engineers are skeptical, arguing that a general slowdown would favor established US companies while hindering the progress of newer players. This skepticism was echoed by researchers at Alibaba Group Holding's Qwen team and former lead researchers at Alibaba's Qwen models. State-owned tabloid Global Times accused Amodei of proposing a "quiet AI Cold War" aimed at widening America's technological lead through technological barriers and regulatory monopolies, while excluding China from global AI governance.

Despite resistance in China and from the US administration, Chinese AI players are accelerating their development timelines. Zhipu AI, a prominent Chinese AI company, has raised substantial funds to advance its next-generation GLM model and a fully self-training system, while researchers from the Shanghai AI Lab and ByteDance have outlined a five-stage roadmap towards autonomous model improvement.