- The September 25, 2026 Washington summit between President Trump and President Xi ended without a binding AI agreement, with US Trade Representative Jamieson Greer confirming “no specific new AI developments” from the meeting.
- Trump stated he intends to “leave it exactly where it is” on advanced AI regulation, citing the Department of Justice as the primary guardrail, while Xi publicly called for AI to remain “under human control.”
- The only concrete AI outcome was an agreement to continue an “AI dialogue” and discuss an incident-notification mechanism, which a US official compared to the Cold War-era hotline between Washington and Moscow.
Three days of pageantry, a White House state dinner with top tech executives and no binding AI agreement: the September 25, 2026 Washington summit between President Trump and President Xi Jinping ended much as it began, with the two largest AI powers publicly committed to competition and privately committed to very little else. US Trade Representative Jamieson Greer confirmed at the close of the visit that while an “AI dialogue” had been established, there were “no specific new AI developments” from the meeting.
What the Summit Actually Produced
The two sides agreed to continue the newly established “AI dialogue” and to discuss an “incident-notification mechanism,” intended as a communication channel for potential AI-related risks. A US official compared it to the Cold War-era hotline between Washington and Moscow, which frames the ambition accurately: de-escalation management, not arms control. Representative Ro Khanna was among the critics who said the outcome failed to meet the moment. No binding rules, no formal framework and no agreed pause in AI development emerged from the talks.
The summit’s guest list telegraphed higher ambitions. Executives from OpenAI, Meta, Nvidia and Google attended Thursday’s state dinner, and calls for a safety framework circulated in the run-up to the meeting. None of it translated into diplomatic language by Friday.
Two Incompatible Positions
The gap between the two leaders on AI governance was wide enough that the absence of a deal was arguably the predictable result. Trump maintained throughout the summit that he intends to “leave it exactly where it is,” using the term “Super Intelligence (SI)” for advanced AI systems and naming the Department of Justice as the primary guardrail. That position reflects a view shared by top Republicans and some in the tech industry: that regulation at this stage would blunt American competitiveness and hand China an opening.
Xi’s public remarks took a different register. Speaking at the White House, he said both countries “have the capability and responsibility to develop and manage AI for good, and ensure that the development of AI is always under human control and serves the well-being of the people.” China has moved to regulate AI domestically in recent months, but its framework centres on domestic content control and data governance, not international limits on development. The gap between the two positions, one rejecting external constraints, the other focused on sovereign control, left little room for a shared governance architecture.
China’s Parallel Moves
While the bilateral talks stalled, Beijing has been building its own multilateral infrastructure. China’s “Global AI Governance Initiative” and its role in establishing the World Artificial Intelligence Cooperation Organization, which brought together 29 countries in July, give it an alternative forum to advance its preferred norms. That effort complicates any US-led push for a unified international framework, particularly across the Global South, where Xi has actively promoted China’s state-led governance model as an alternative to Western approaches.
Countries outside the two blocs face a straightforward problem: the AI systems and standards shaping their digital infrastructure will be built in Washington or Beijing, with limited input from anyone else. Beijing’s pitch to the developing world has focused on strategic flexibility and non-interference, a contrast to Washington’s demands for alignment on export controls and geopolitical positioning.
The Export Control Backdrop
A decade of US chip export controls aimed at constraining China’s AI industry sits behind the summit’s atmosphere, and it shows. China’s domestic AI computing capacity is estimated at between 1% and 4% of the United States’, making continued access to US-origin semiconductors a live strategic concern for Beijing. That dependency gives Washington leverage but also gives China a concrete grievance that makes broad cooperation harder to negotiate.
The summit closed without a formal press conference. Trump has consistently framed AI competition with China as the reason to resist safety constraints, and the Washington meeting did nothing to alter that calculus. What the two governments are left with is a dialogue mechanism and an unresolved arms race, which may be exactly what both sides intended to leave with.
