> ## Content Index > Fetch the complete content index at: https://beirutwire.com/llms.txt > Use this file to discover other available public pages before exploring further. # The U.S. and China Open Formal Talks on AI Risks Ahead of Trump-Xi Summit - URL: https://beirutwire.com/the-u-s-and-china-open-formal-talks-on-ai-risks-ahead-of-trump-xi-summit/ - Published: 2026-09-21T05:15:22.000Z - Updated: 2026-09-21T05:15:22.000Z - Description: The U.S. and China have agreed to begin formal discussions on AI risks ahead of Donald Trump’s meeting with Xi Jinping in Washington. - Author: Beirut Wire - Tags: Technology The United States and China have agreed to begin a formal dialogue on artificial intelligence risks, creating a new channel between the world’s two leading AI powers just days before President Donald Trump meets Chinese President Xi Jinping in Washington. The agreement emerged from talks in New York between U.S. Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng. Bessent proposed a new notification mechanism under which Washington and Beijing could alert one another about major AI incidents with national-security implications, including dangerous system failures, cyber incidents or other events that could escalate tensions. The significance is difficult to overstate. The United States and China are not simply competitors in artificial intelligence. They are the two countries most capable of pushing frontier AI forward, and increasingly, both sides recognize that the risks created by more powerful models may eventually require some degree of coordination even between geopolitical rivals. That does not mean the AI race is slowing. Washington still views maintaining an AI lead over China as a core economic and national-security priority. Trump has repeatedly rejected calls for sweeping new limits on AI development, arguing that excessive regulation could allow China to catch or surpass the United States. China, meanwhile, is rapidly closing parts of the technological gap. Chinese companies including DeepSeek and Moonshot AI have developed increasingly competitive models, while Beijing has invested heavily in domestic chips, data centers and open-source AI as it seeks to reduce dependence on American technology. U.S. officials have also accused some Chinese companies of using distillation to copy capabilities from American models, allegations Beijing has rejected. That makes the new dialogue unusual. The two countries remain locked in a contest over chips, computing power, model capabilities and the broader economic value of AI. Yet they are now simultaneously discussing how to manage the risks created by that competition. The closest historical comparison may be nuclear arms control. During the Cold War, Washington and Moscow continued building nuclear arsenals while also establishing hotlines, notification systems and agreements designed to reduce the risk of accidental escalation. Some AI researchers now argue that the United States and China may eventually need something similar for advanced artificial intelligence. Experts from both countries have already proposed measures resembling nuclear-era safeguards, including mandatory notification of dangerous AI incidents, shared testing standards and communication channels for crises involving autonomous systems. Those proposals have not yet been formally adopted by either government. The immediate U.S. proposal is considerably narrower. Rather than trying to establish a global regulatory system, Washington is initially focused on creating a mechanism through which the two governments could notify each other when an AI incident creates a serious national-security risk. That could matter in several scenarios. An autonomous AI system involved in a cyberattack could be mistaken for deliberate state action. An AI-controlled military system could malfunction. A model could be used to attack critical infrastructure. Or an advanced AI agent could behave unpredictably in a way that affects another country. In any of those cases, the ability of Washington and Beijing to communicate quickly could help prevent a technological incident from becoming a geopolitical crisis. The talks come at a moment when concern about frontier AI is rising rapidly inside the industry itself. Anthropic CEO Dario Amodei has called for AI companies to slow the race toward increasingly powerful systems and has pushed for greater international cooperation. OpenAI CEO Sam Altman has similarly argued that even a limited agreement between the U.S. and China on testing and shared safety standards could represent a major breakthrough. Trump, however, has taken a much more aggressive position on domestic AI development. He recently dismissed many concerns about AI guardrails, arguing that America must continue moving quickly or risk surrendering technological leadership to China. That creates an important distinction. The Trump administration does not appear to be advocating a slowdown in the American AI industry. Instead, it is attempting to separate two issues: maintaining maximum technological competitiveness at home while establishing limited international mechanisms to prevent catastrophic misunderstandings or accidents. China appears to be approaching the issue from a similarly strategic perspective. Beijing has publicly emphasized AI safety and regulation while simultaneously rejecting what it views as American attempts to restrict China's access to advanced chips and models. Chinese officials argue that global AI governance should not become a justification for preserving U.S. technological dominance. Mutual distrust therefore remains enormous. The United States worries that China could use cooperation to gain access to sensitive technological information. China believes Washington is using export controls and national-security rules to slow Chinese development. Both sides also accuse the other of using AI in ways that could threaten critical infrastructure and national security. That means expectations for the Trump-Xi meeting should remain limited. Reuters reports that AI will be one of several major issues on the agenda when Trump hosts Xi in Washington, alongside trade, Taiwan, critical minerals and the Middle East. The summit is scheduled for September 24. But even beginning a formal AI dialogue would represent a notable change. Until now, the relationship between the United States and China on artificial intelligence has largely been defined by competition: chip restrictions, export controls, model development and efforts to build rival technological ecosystems. The new talks introduce another possibility. The two countries may increasingly treat advanced AI not only as something to win, but as something powerful enough that even rivals need rules for what happens when it goes wrong. That does not end the AI race. If anything, both sides appear determined to accelerate it. But it may mark the beginning of something the nuclear age eventually forced on previous superpowers: recognition that competition and cooperation sometimes have to exist at the same time.