US and China Push for September AI Talks Ahead of Xi's Visit to US... "Bessent to Attend"

Amid ongoing technological and security competition between the U.S. and China over the artificial intelligence industry, the two countries are reportedly pushing for intergovernmental talks on AI in September.

Citing multiple anonymous sources, Reuters reported on the 21st (local time) that the AI talks, agreed upon during the U.S.-China summit last May, could take place before Chinese President Xi Jinping's visit to the United States on September 24.

Sources said U.S. Treasury Secretary Scott Bessent is scheduled to lead the U.S. delegation.

However, they explained that since preparations for the talks are still in the early stages, other participants from both sides, as well as the venue and agenda, remain under review.

The exact date is also known not to have been finalized yet.

Relevant agencies from both countries, including the White House, the State Department, and China's National Development and Reform Commission (NDRC), declined to comment on Reuters' requests.

Meanwhile, Secretary Bessent avoided a direct answer when asked about the AI talks during an interview with Fox Business on this day, but expressed concerns regarding intellectual property rights and stated that the same standards should be applied to AI models.

He added, "We are finding 'watermarks' from large U.S. language models (LLMs) in many Chinese models, which is unacceptable," and stated, "We will be looking into this in the coming days or weeks."

One source reported that the Chinese side will continue internal discussions until next month before deciding on the departments and officials who will attend the talks.

The source also explained that China hopes the upcoming talks will serve as a venue for practical communication and technical discussions rather than political negotiations.

However, some observers speculate that since these talks will be the first held under the U.S. administration of Donald Trump, they are more likely to seek alignment on basic terms such as the definition of "frontier AI models" rather than resolving major issues.

During the previous U.S. administration of Joe Biden, the two countries held working-level talks in 2024 and agreed that humans, not AI, should retain decision-making power over the use of nuclear weapons, but there have been no official AI-related talks under the Trump administration yet.

Both the U.S. and China are deliberating on ways to mitigate risks posed by each other's frontier-level AI models.

The United States regulates the export of advanced AI chips to China, and is reportedly considering restricting domestic companies from using Chinese AI models following the release of low-cost, high-performance AI model "Kimi K3" by Chinese startup Moonshot.

Amid concerns over potential hacking using U.S. firm Anthropic's Mythos model, China is exploring ways to restrict foreign access to its own advanced AI models.

(Photo: AP, Yonhap News)
※ Please note: This article was translated by AI and may contain errors.