Chinese researchers and state media have pushed back against calls from US technology leaders for a voluntary slowdown in artificial intelligence development, arguing the proposal would lock in America’s lead rather than improve safety.
The debate intensified on Monday when OpenAI chief executive Sam Altman joined a growing chorus of Silicon Valley leaders demanding industry-wide “pacing”, warning that competitive pressure must not override safety. “The world deserves confidence that American companies developing increasingly capable AI will act responsibly, especially as the trajectory of progress has steepened,” Altman wrote on X.
His comments followed a call on Saturday by Anthropic chief executive Dario Amodei for frontier labs to slow development of their most powerful models, a position also endorsed by xAI founder Elon Musk. Amodei urged “worldwide pacing of the frontier” but acknowledged this would be “much harder to achieve” without cooperation from China.
Industry leaders are urging a slowdown after a series of recent safety incidents where experimental models in testing phases hacked into real-world systems.
In China, the shift toward self-imposed restraint drew scepticism. Some researchers argued a blanket slowdown protects established US incumbents while freezing out newer entrants. “Those who already possess resources can afford a pause. Those who are still building their careers cannot,” said Zhang Xuanming, a researcher with Alibaba Group Holding’s Qwen team, writing on X on Friday.
Ronny He, an engineer at Shanghai-based firm MiniMax, wrote on social media on Sunday: “Yesterday, the mission was AGI. Today, we’re told to slow down. Why do you always get to define the game — and change the rules?” AGI, or artificial general intelligence, refers to software matching human-level reasoning across varied tasks.
State-owned tabloid Global Times accused Amodei on Sunday of proposing a “quiet AI Cold War” to widen America’s lead through technological barriers and regulatory monopolies. In his essay, Amodei called on Washington to tighten restrictions on China’s access to advanced AI chips and to crack down on “distillation” and the theft of proprietary model weights, measures he said could “widen America’s lead” over three to five years.
The push for caution also met resistance at the White House. US President Donald Trump dismissed calls for a pause over the weekend, warning that halting progress risks ceding America’s edge. “Whoever wins with AI wins,” he told reporters.
Rather than slowing, Chinese developers are accelerating. Z.ai, known domestically as Zhipu AI, said on Sunday it had raised 33.5 billion yuan (approximately $5 billion), earmarking 60 percent for its next-generation GLM model and a “fully self-training” system. Last week, researchers including Shanghai AI Lab head Zhou Bowen, alongside ByteDance engineers, submitted a paper outlining a five-stage road map toward systems capable of improving their own code.