Trump-Xi summit holds first US-China AI talks with no deal reached

By Splitfeed EditorialLast updated Sep 26, 2026, 20:18 UTC

Between September 22 and 25, 2026, Donald Trump and Xi Jinping held a summit that included the first formal government-to-government dialogue on artificial intelligence between the United States and China. China confirmed the AI talks took place alongside the eighth round of China-US economic and trade consultations. Scott Bessent announced that the US proposed an AI incident notification mechanism to China, and both sides agreed to establish an ongoing AI dialogue. The summit produced a tariff agreement and two new deadlines, but closed without any binding AI safety deal.

The significance of the talks lies less in what was agreed and more in the precedent they set. For the first time, AI safety has a formal channel between the two governments most central to shaping the technology's trajectory. At the same time, CNBC reported that neither side showed willingness to slow its own AI development, which raises the question of whether a dialogue without constraints amounts to much. The proposed notification mechanism for AI incidents affecting national security is concrete but narrow, and what counts as an AI incident, who decides, and what follows from a notification remain entirely open questions.

The anti-AI camp is treating the talks with outright suspicion. Robert Maxwell argued the US and China share an interest in global surveillance rather than in protecting citizens. Sally Deal pointed to Trump's personal holdings in AI infrastructure and semiconductor stocks as a reason to doubt he would support any deal that curbed the industry. Tom Hearden and the Social Media Lab account framed AI risk in catastrophic terms, with one describing it as nuclear-level and existential, suggesting the talks are inadequate to the scale of the problem. The pro-AI camp has not yet responded publicly, though a story of this kind would typically draw arguments that formal diplomacy between the two leading AI powers is a net positive and that even a limited dialogue creates infrastructure for future agreements. The middle ground camp is cautiously constructive: James Cameron called the dialogue and the notification mechanism sensible and potentially productive, while also noting that neither country is likely to accept anything that restrains its own competitive position. Tech Policy Press added a layer of skepticism by observing that no one can independently verify what AI companies claim, which complicates any safety framework that relies on self-reporting.

The details most worth watching are whether the notification mechanism gets formally defined and adopted, what form the AI dialogue takes when it next convenes, and whether the two new deadlines cited from the summit produce any substantive rules or simply reset the clock for another round of talks.