US AI slowdown calls collide with China competition

Calls to slow frontier AI development are gaining support among several US AI leaders, but they face resistance from the US government and China.

After Anthropic researcher Jacob Coxon resigned and warned that unchecked progress could create an immediate existential risk, Anthropic head Dario Amodei, OpenAI CEO Sam Altman, and Grok founder Elon Musk backed a slower approach. Amodei said that gaining an extra year or two before models reached critical capability could help advance alignment and reduce risk.

Amodei proposed a three-part plan and said Anthropic would begin by giving third-party evaluators permanent, employee-level access to its systems. The evaluators would check safety measures, report incidents, and assess model alignment during training. Altman said AI regulations and standards were not ready to support continued acceleration.

President Donald Trump rejected a slowdown, arguing that it could let China catch up. He said the United States was leading China and that whoever wins AI wins. Beijing called the proposals fearmongering from a Cold War playbook. Chinese Foreign Ministry spokesperson Guo Jiakun said confrontation and competition would disrupt global AI governance.

The source also compares the situation with the nuclear arms race and cites physicist Leó Szilárd’s 1950 discussion of a cobalt bomb. It argues that company-level controls will struggle unless China, the United States, and major AI labs coordinate, while enterprises still need governance over what AI agents can access, change, and decide.