Tensions between the United States and China over artificial intelligence (AI) have intensified ahead of a scheduled summit between US President Donald Trump and Chinese President Xi Jinping, reflecting broader geopolitical rivalry and differing approaches to AI governance and safety.
In recent weeks, unofficial talks have taken place between Chinese think-tanks and technology companies and their American counterparts as both sides seek to address the risks associated with increasingly powerful and potentially uncontrollable AI systems. Participants in these discussions acknowledge shared concerns over AI’s dangers but say deepening technological competition complicates efforts to establish joint safety measures or “guardrails.”
The dispute was further heightened following an essay published by Dario Amodei, CEO of the US AI company Anthropic, calling for a slowdown in AI development among leading American labs. Amodei also advocated for measures that would enable the US to retain its technological edge over China, including restricting China’s access to advanced semiconductor chips and challenging Chinese entities that develop AI models by training on outputs generated by US rivals—a practice known as distillation.
Chinese experts involved in the dialogue expressed disappointment with Amodei’s stance, viewing it as narrowing the possibility of cooperation. Xiao Qian, vice-dean of the Institute for AI International Governance at Tsinghua University, said linking AI safety to national security concerns made collaboration “even harder” and more complicated.
The situation drew criticism from Chinese media aligned with the Communist Party, which characterized Amodei’s claims as reducing AI safety to a geopolitical contest rather than a global challenge requiring cooperation. An unnamed Chinese expert cited by one tabloid asserted that commercial interests partly motivated Amodei’s assertions and warned against framing AI leadership as a “grave danger” specifically to the US and the world.
Ahead of the summit set for September 24, sources familiar with US government deliberations have expressed low expectations for any significant progress on AI cooperation between the two leaders. Although China reported that an AI dialogue had been agreed upon earlier in the year, formal discussions have yet to commence.
Another contentious issue is the US allegation that Chinese AI labs employ distillation to develop competing models quickly and at lower cost, a practice the US government sees as undermining fair technology development. During a July meeting of the Asia-Pacific Economic Cooperation forum, US Commerce Department official Bill Guidera raised these concerns with China’s Cyberspace Administration. Chinese officials have rejected the claims as baseless and insisted that domestic companies’ independent innovations are overlooked by such accusations.
The US National Security Agency and other agencies publicly accused several Chinese firms, including DeepSeek and Alibaba, of engaging in distillation activities, suggesting such efforts were likely carried out with government knowledge. Chinese specialists argue that these allegations inflame political tensions and impede consensus on AI cooperation.
Observers note that US and Chinese authorities prioritize different AI risks. US discussions emphasize existential threats posed by unchecked AI advancement, while China focuses on immediate issues like cybersecurity and controlling the spread of politically sensitive content. A person familiar with the Chinese regulatory outlook commented that strong AI supervision is not currently a priority since China is regarded as trailing in the technology.
Within Chinese AI industries, conversations about catastrophic AI risks remain limited. An investor involved with two large Chinese language model groups remarked that AI safety is rarely addressed seriously among investors and executives, with recent dialogues mostly informal and less extensive than in the US. The emphasis in China remains heavily on advancing AI capabilities.
Nonetheless, AI risks are becoming more prominent on China’s political agenda. On September 22, the Ministry of State Security issued a comprehensive warning highlighting AI-related dangers such as cyberattacks and political destabilization. A source close to Chinese regulators interpreted this as a signal to the US that a platform exists for AI discussions during the upcoming summit, demonstrating Beijing’s recognition of AI threats.
As the Trump-Xi meeting approaches, the underlying strategic competition between the two countries continues to pose significant challenges for meaningful cooperation on AI regulation and shared safety concerns.