The US-China race and rivalry over artificial intelligence

In April the White House science adviser, Michael Kratsios, said foreign entities mainly in China were copying frontier models from US companies.^[2] In July Moonshot AI's Kimi K3 model stunned developers by appearing to rival top American systems at a fraction of the cost,^[3] and the White House accused the company of gaining access to restricted Nvidia systems and possibly using Anthropic's technology.^[6] The administration signaled it might ban advanced Chinese models,^[7] Treasury Secretary Scott Bessent warned of sanctions over intellectual property theft,^[8] and Chinese officials discussed limiting foreign access to their own best models.^[9] Chinese President Xi Jinping cast Beijing as a partner for AI cooperation.^[10][11]

Anthropic's chief executive, Dario Amodei, urged tighter US restrictions on chips and technology to China, which China's foreign ministry called "fearmongering."^[12][13] Ahead of Xi's state visit, Bessent said the US was open to discussing shared risks,^[14] and US officials proposed an AI safety notification mechanism in talks with Chinese counterparts.^[15] At the summit Xi said both countries had an obligation to manage AI as "leading nations,"^[16] and the two agreed to the dialogue.^[5] The summits are followed in their own arc. Afterward Bloomberg reported that China's top models trailed US systems by just 3% on key benchmarks,^[4] and analysts said China was winning ground in the Global South with cheap open-weight models.^[17]