US and China discuss new AI dialogue channel for security concerns
US and Chinese officials have discussed establishing a bilateral AI dialogue to exchange information on AI-related incidents with potential national security implications.
The United States and China have discussed creating a dedicated communication channel for artificial intelligence issues, following talks between US Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng in New York on 20 September. Bessent said the proposed mechanism, referred to as the US-China AI Dialogue, would provide a channel for discussing shared AI-related concerns.
The proposed channel would focus in particular on incidents involving national security. Bessent said the two countries were seeking greater transparency and a way to notify each other about potential incidents or situations in which AI could create national security concerns. The proposal is expected to be discussed further around the planned 24 September meeting between US President Donald Trump and Chinese President Xi Jinping.
The initiative comes as the two countries continue to compete over AI development while also discussing risks that could affect both sides. Earlier in September, Bessent said the United States was open to discussing shared AI risks with China, including issues involving different types of AI models.
The precise structure of the proposed dialogue remains limited. Reporting by Politico indicates that the mechanism under discussion would initially be a direct communication channel between Bessent and He Lifeng rather than a formal expert body. Chinese officials have also not publicly provided details confirming the mechanism’s structure.
The discussions are taking place alongside broader US-China negotiations on trade and economic issues. The New York meeting between Bessent and He Lifeng was held ahead of the leaders’ summit in Washington and included US Trade Representative Jamieson Greer and Chinese officials.
The proposed AI channel would therefore focus on communication and incident notification rather than establishing a common regulatory framework for AI. Its development could provide a direct government-to-government mechanism for addressing specific AI-related security concerns, while leaving broader differences between the US and Chinese approaches to AI governance unresolved.
