The United States and China, the world's two leading AI powers, hold fundamentally different perspectives on how to regulate artificial intelligence. This divergence is poised to shape global norms, trade, and competition for years to come. While both nations acknowledge the need for AI guardrails, their primary concerns and motivations differ substantially.
US tech leaders, such as Anthropic CEO Dario Amodei, have increasingly emphasized the potential for AI to pose existential risks to humanity, advocating for a slowdown in AI development and proposing safeguards that include restricting China's access to advanced AI technology. This stance has been met with skepticism and strong opposition from China.
China, on the other hand, views such calls as a "Cold War playbook" aimed at hindering its technological progress and maintaining US dominance. Beijing's primary anxieties regarding AI are centered on its potential to disrupt domestic social stability and threaten the Chinese Communist Party's grip on power, rather than hypothetical existential risks. Chinese policymakers also suspect that US companies' calls for AI safety are instrumental, possibly aimed at inflating their own value or slowing down competitors.
Despite these differences, preliminary negotiations are reportedly underway between the two countries ahead of a planned meeting between Chinese leader Xi Jinping and US President Donald Trump. Both governments share concerns about cybersecurity threats, the misuse of powerful AI models, and unexpected AI system behaviors. However, the deep mutual suspicion and differing priorities suggest that any progress in cooperation will likely be incremental, focusing on narrowly scoped issues with credible verification mechanisms.
Security experts from both nations have proposed nuclear-style safeguards for AI risks, including "red lines" around nuclear systems, human-only military decisions, and an AI incident hotline. These recommendations come amidst a growing AI rivalry that spans from chips and models to strategic stability, highlighting the urgent need for mechanisms to prevent AI from escalating conflicts, such as an AI system interfering with nuclear command networks.
