As AI leaders warn of catastrophe, US and China shun slowdown calls

Calls from prominent AI executives and researchers for a global slowdown in advanced AI development have met resistance from the two countries that drive most frontier work: the United States and China. As Presidents Donald Trump and Xi Jinping prepare to meet in Washington, observers expect only limited cooperation on safety measures rather than coordinated pauses or strict international restraints. Observers and experts say US-China agreement is crucial because the two nations host nearly all frontier AI labs and most widely used models. Alvin Wang Graylin of the Asia Society Policy Institute argued that guardrails agreed by both sides would be a nearly essential step for safer development, but analysts warn that deep mistrust and divergent assessments of primary risks hinder cooperation. Officials in Washington have downplayed external controls: President Trump has called AI fears a 'conspiracy' and told the United Nations General Assembly the US would oppose any 'globalist scheme' to control the technology. Beijing has issued its third AI security framework this month and pushed back against Silicon Valley warnings of existential danger, framing risks primarily around cybersecurity, misuse, and threats to the ruling party's authority. Chinese Minister of State Security Chen Yixin warned of 'hostile forces' exploiting AI to spread political misinformation. Within the AI sector there is no consensus on the proper regulatory approach. Leaders such as Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman have urged a slowdown and stronger oversight, while figures including Nvidia CEO Jensen Huang and Meta CEO Mark Zuckerberg have minimized the need for external supervision. Some researchers, like David Krueger, have called for an immediate and indefinite international moratorium on frontier AI, and commentators such as Otto Barten have urged explicit red lines to prevent recursive self-improvement in AI systems. So far, the only bilateral proposals publicly mentioned are establishing a US-China dialogue and a notification mechanism to report national security threats. Experts say these steps could be useful initial measures but that they fall short of deciding which AI capabilities or uses should be off-limits to prevent the most severe risks. Analysts caution that if safety discussions are perceived as a cover for broader technology competition — for example, as a way to impose export controls — Beijing may be reluctant to engage in substantive cooperation.