President Donald Trump told the Asia-Pacific Economic Cooperation meeting in Shenzhen on November 19 that the United States opposes any form of "globalist regulation" of artificial intelligence. Chinese leader Xi Jinping, speaking at the same gathering, called for US-China cooperation to keep AI under human control. The two positions sharpen a split that has defined AI governance for the past year, and they set the stage for a bilateral meeting held later the same day.
What Trump Said
Trump described AI as potentially "super intelligence" that "we have to watch," and said it is "almost completely good" but has harmful elements that must be managed carefully. He drew a firm line on who should set the rules. The United States, he said, would not allow "globalist regulation" or anything that would "stifle" the technology.
The remarks continue a pattern. Trump has repeatedly rejected international AI governance, arguing that it could hold the United States back in its race with China. His position favors national competitiveness and minimal oversight over multilateral agreements, a stance that puts Washington at odds with several allies who want shared standards for frontier systems.
Xi's Counter-Message
Xi took a different tone. According to Chinese state media, he called on the two countries to "strengthen communication and coordination to guard against artificial intelligence safety risks" and to ensure that the technology "always remains under human control." He also said at APEC that leaders should deepen cooperation on AI and the digital economy, though he gave no specifics in the coverage available.
The contrast is stark. Washington rejects shared rules, while Beijing presents itself as a partner in safety. Both leaders, however, are describing the same underlying concern, which is that powerful systems must remain controllable. The disagreement is about who should define control and through what institutions.
A Channel for AI Incidents
The most concrete element on the table is a communication channel. In September, Trump and Xi agreed to create a channel for urgent communication about AI incidents, with the details left to be worked out at the talks. Such a mechanism would let officials raise alarms quickly if an AI-related event threatened national security or public safety.
Shared concerns cited in coverage include attacks on critical infrastructure and the misuse of AI to help build biological weapons. Military use of AI and mass surveillance are potential flashpoints, which means any channel would need to address areas where the two countries have very different interests. A hotline is only as useful as the agreed definition of what counts as an incident.
Wider Concerns and Export Controls
Canadian Prime Minister Mark Carney used the same forum to warn that the lack of global AI governance poses an "enormous risk." His comment reflects a view held by many middle powers that they will have little say in rules written by the United States and China alone.
The backdrop also includes trade. The United States recently announced tighter export restrictions on advanced chips to China. Chips are the physical base of frontier AI, so export policy is effectively a form of AI governance, even when it is framed as trade or national security. Companies that design or manufacture those chips will feel the rules before most users do.
Why It Matters
For companies, the speech signals that the federal government is unlikely to back broad international rules in the near term. That shifts the burden onto domestic law and voluntary commitments, and it means firms operating across borders may face a patchwork of obligations that differ by jurisdiction. Compliance teams will need to plan for rules that do not converge.
For governments outside the two largest AI powers, the message is that they will need to build their own frameworks or join coalitions that can speak with weight. The incident channel is the most practical outcome to watch. Even if broader governance stalls, a working line of communication between Washington and Beijing could matter in a crisis, especially if AI-enabled cyberattacks or military misreads occur.
What to Watch
Three developments will show what the Shenzhen meeting produced. The first is whether the incident communication channel is formally established, with named contacts and procedures. The second is whether the export control posture changes after the bilateral talks. The third is whether other governments move to set rules for frontier systems in response to the stalled global effort.
For now, the clearest takeaway is that the world's two biggest AI powers agree on the risks in principle but disagree on whether those risks should be managed through shared rules.
