US and China Schedule September Talks on Artificial Intelligence

The United States and China are preparing their first official talks on artificial intelligence since Donald Trump returned to power. According to Reuters, citing people familiar with the matter, the dialogue is set to take place in September, ahead of President Xi Jinping's planned September 24 visit to the United States.

Preparations are still at an early stage. The exact date, venue, and the full composition of the delegations on both sides have yet to be determined. No official body, neither the White House, the State Department, nor Chinese ministries, has publicly confirmed the details so far, and the information comes from anonymous sources cited by Reuters.

Shadow of technology theft accusations

The talks are set to begin against a backdrop of a growing dispute over whether Chinese companies are illegally copying American language models. Treasury Secretary Bessent spoke bluntly about traces of other people's work showing up in Chinese systems. In recent weeks, Washington has repeatedly accused Chinese firms, including Moonshot AI, the maker of the Kimi K3 model, of using stolen data and illegally imported chips subject to export embargoes.

We are finding watermarks of our American large language models in many Chinese models, and that is unacceptable - Scott Bessent, US Treasury Secretary

US Trade Representative Greer has also weighed in on the dispute, announcing a close look at how China is developing its artificial intelligence to ensure a level playing field for American companies. These statements show that the September dialogue will not be a purely technical meeting of experts, but part of a broader contest for technological and trade advantage between the two powers.

Beijing's priorities

The Chinese side has declared its readiness to cooperate on implementing the agreements from the May Trump-Xi summit, describing artificial intelligence as one of the biggest challenges facing the international community. Above all, though, Beijing wants to learn about American regulatory plans for controlling the release of further advanced models, and to find out whether Washington will impose restrictions on Chinese open-weight models.

Chinese officials also have questions about specific American systems, including Anthropic's Mythos model, which in recent weeks has been the subject of separate warnings from the UK's AI Safety Institute about security vulnerabilities. Beijing is also considering imposing its own restrictions on foreign access to the most advanced Chinese models, suggesting that AI export controls could soon run in both directions.

First such dialogue since Biden

This will be the first official bilateral AI dialogue since Trump took office. Under the Biden administration, working-level talks were held in Geneva in 2024, but they had a much narrower scope and lower-ranking participants than the meeting now being planned with the head of the US Treasury Department.

The backdrop to the dispute is tense. In recent months, Washington has accused Chinese firms of stealing models through so-called distillation, threatening sanctions, while the Trump administration has argued internally over a ban on Chinese AI models, a move opposed by roughly 200 American startups that rely on cheap, open Chinese solutions. At the same time, the United States has cut off Europe's access to Anthropic's most powerful models, showing that Washington treats AI export controls as a tool of global policy, not just rivalry with Beijing.

What this means for Poland and Europe

For Polish companies and institutions using open Chinese models such as Kimi K3 or the DeepSeek models, the outcome of the September talks could have practical consequences. If Washington decides to formally restrict access to Chinese models, or Beijing imposes its own export restrictions on its best systems, European users could find themselves caught between two competing regimes of technology control, much as they were already cut off from the most powerful American models.

The talks will also test whether the two powers can work out even a basic framework of safety rules for the most advanced systems before trade and technology rivalry turns into an open regulatory conflict. For now, neither side has indicated whether the meeting will produce anything more than a pledge to keep talking.