NEW YORK β The United States and China have opened discussions on a formal channel for addressing artificial-intelligence risks, with senior officials from both countries meeting in New York on September 20-21 ahead of President Donald Trump's planned meeting with Chinese President Xi Jinping, as Washington and Beijing seek limited cooperation while continuing their wider technological competition.
The discussions involve U.S. Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng and focus on creating a mechanism through which the two governments could communicate about potentially dangerous AI incidents. Areas under discussion include cyberattacks, biological misuse, autonomous systems and other situations in which advanced AI could create risks that cross national borders.
The proposed arrangement is significant because the United States and China remain the world's two largest AI powers and are competing heavily over advanced computing, semiconductor technology, AI models and data infrastructure. The rivalry has also produced restrictions on technology exports and continuing accusations over the transfer and development of AI capabilities.
Despite those disagreements, officials on both sides have recognised that certain AI-related risks cannot easily be contained within one country. A serious cyber incident, misuse of an advanced model or an AI-related threat to critical infrastructure could affect organisations and systems outside the country where it originates.
The discussions are therefore focused on communication rather than a broad agreement over how artificial intelligence should be developed. The two governments continue to have major differences over technology policy, national security and access to advanced computing equipment.
The talks also coincide with Trump's decision to encourage rapid AI development inside the United States. On September 21, Trump said on Truth Social that he did not intend to restrict the growth of artificial intelligence and argued that the technology could become more significant than previous industrial and technological revolutions. The president has also announced plans for an βAI Forceβ and said he intends to appoint an AI czar, although details of the proposed structure and its responsibilities remain limited.
China, meanwhile, continues to invest heavily in domestic AI development. Chinese companies including DeepSeek and Moonshot AI have become major participants in the global AI race, while Beijing has sought to strengthen its ability to develop advanced models and reduce dependence on foreign technology.
The proposed U.S.-China communication channel does not remove those strategic tensions. Instead, it represents a narrower area where the two countries are exploring practical communication to reduce the possibility that an AI-related incident could escalate because of a lack of direct contact.
The timing is also important because Xi is scheduled to visit the United States from September 23 to 25 at Trump's invitation. The visit is expected to cover a broad range of issues involving trade, technology, security and relations between the two countries.
As the two governments prepare for that high-level meeting, AI has emerged as one of the areas where competition and cooperation are developing simultaneously. Washington and Beijing remain rivals in the race for technological leadership, but their latest discussions show that both sides also recognise the need for channels capable of dealing with risks created by increasingly powerful AI systems.
