China rejects US AI slowdown calls as bid to lock in Western advantage

Two years ago, Boris But, who runs an education company in China, had a moment of clarity: his business would either adopt artificial intelligence or become irrelevant. His clients—affluent parents seeking to place their children in overseas schools—were arriving at his office armed with information and opinions pulled not from competing firms but from chatbots. He built his own AI study tool. "If you can't beat them, join them," he concluded.

This small decision reflects a much larger one playing out across China. The country's AI development now ranks second globally, behind only the United States, with the two superpowers engaged in an escalating competition as each new model's capabilities raise the stakes. American frontier models remain ahead by most measures, but the gap is narrowing. Yet in recent weeks, the tone from Silicon Valley has shifted toward alarm. Researchers and executives have warned repeatedly of catastrophic risks from losing control of the most advanced systems. Last Saturday, Dario Amodei, the chief executive of Anthropic, called for a "worldwide pacing of the frontier"—a slowdown he said would require cooperation with China.

China is not interested in slowing down on those terms. For Beijing, AI represents a core sovereignty capability, something too central to national power to compromise. Lian Jye Su, an analyst at the technology research firm Omdia, put it plainly: China sees these calls for restraint as a strategy to lock in American advantage. Amodei himself made this explicit, arguing that maintaining democracies' AI lead over autocracies was essential and advocating for continued export controls on advanced chips—measures that some argue already amount to a forced, one-sided slowdown imposed on China. When China's foreign ministry responded to Amodei's essay, it rejected what it called "fearmongering, confrontation and vicious competition."

But the picture is more complicated than a simple rejection. This week, China released the third version of its AI safety governance framework, a document that maps the AI risk landscape and outlines responses. The updates address emerging concerns: risks from autonomous agents, cybersecurity threats from frontier models, and recursive self-improvement—the hypothetical scenario where AI becomes sophisticated enough to improve itself. Gabriel Wagner, a researcher at Concordia, a Beijing-based AI safety consultancy, noted something striking: "In many ways, there is more common ground than people might assume between what Amodei is saying and what the Chinese government is saying." This contrasts sharply with U.S. President Donald Trump, who this week dismissed fears of AI-driven human extinction as a "hoax." Trump is scheduled to meet Xi Jinping in Washington next week, with AI expected on the agenda, though analysts are skeptical any major agreement will emerge.

China's approach reflects its political structure. The Communist Party, acutely aware of threats to its control, has ensured AI develops within state-set boundaries. Centralized authority over critical infrastructure and commerce means China is arguably better positioned than the United States to manage an AI disaster, Su said. Yet Chinese companies conduct less voluntary monitoring and evaluation than their American counterparts, according to Leia Wang of the Carnegie Endowment for International Peace. Neither government appears truly prepared for a loss-of-control scenario. China's regulatory approach, though, is "iterative"—quick to adapt and release new measures. Since 2022, China has maintained an algorithm registry. This year's warnings about autonomous agents suggest a shift from controlling what AI says to controlling what it does. Beijing also issued national regulations aimed at preventing emotional dependency on companion AI systems.

Meanwhile, AI is being actively encouraged across the Chinese economy as a growth driver, a goal enshrined in the latest Five Year Plan. Local governments offer funding and reduced rents to companies meeting these targets. But's education company was recently invited to pitch to a small city three hours from Shanghai that is positioning itself as a tech hub. Across China's vast manufacturing base, factories are being pushed to adopt AI processes. A representative from Black Lake Technologies, an industrial AI platform working with 40,000 factories globally, noted that while safety and controllability are discussed, manufacturers care most about reliability and operational performance.

Daily life in China is already highly digitized, with less public discourse around data privacy than in the West, enabling rapid AI adoption. A-fu, a digital health service featuring AI chatbots trained on clinical studies and modeled after prominent doctors, recently reached 150 million users. Ni Tao, founding editor of the tech website yangtzeer.com, observed that most Chinese treat AI as a practical everyday tool rather than an abstract threat. "There is real anxiety about jobs and misinformation," he said, "but it coexists with a strong belief that China must master AI rather than slow down." That conviction—that falling behind is the greater risk—shapes everything Beijing does.