The United States proposed that China establish a bilateral mechanism for notifying each other about artificial intelligence-related incidents whose consequences could reach the level of national security, during economic talks held by officials from the two countries in New York on Sunday 20 September, ahead of an expected summit this week between US President Donald Trump and his Chinese counterpart, Xi Jinping.
Bessent proposes mutual notification of AI incidents
US Treasury Secretary Scott Bessent said, following talks with Chinese Vice-Premier He Lifeng, that the two sides had discussed launching a new dialogue between Washington and Beijing on artificial intelligence, with a particular focus on national security concerns and establishing a system to notify each other about incidents and issues that go beyond that threshold.
Bessent said the proposal sought to develop a “shared vision of shared goals and threats”, arguing that greater transparency between the United States and China was necessary as the world’s first- and second-largest powers in artificial intelligence. He added that moving from opacity towards greater transparency represented an important step, as in other activities that transcend national borders.
Beijing considers proposal without taking final position
Beijing’s position on the proposed US mechanism was unclear. An official Chinese statement said the delegations had held dialogue on artificial intelligence-related issues, without announcing details of the notification system or the range of incidents it could cover.
China’s Ministry of Commerce said He Lifeng held talks with Bessent and US Trade Representative Jamieson Greer, and that the two sides had conducted what it described as a candid, in-depth and constructive exchange on implementing understandings reached in previous rounds, strengthening bilateral trade and investment, and dialogue on artificial intelligence.
He Lifeng and China’s chief trade negotiator, Li Chenggang, left the talks’ venue in New York without speaking to the media. Greer said US export controls on advanced AI chips and semiconductor manufacturing equipment had not been included in discussions on the notification mechanism.
Greer added that the New York talks had paved the way for what he described as a summit between Trump and Xi targeted at success, scheduled to take place in Washington on Thursday and Friday, 24 and 25 September. He said the United States and China were the two strongest countries in trade and artificial intelligence, and that their ability to work together remained necessary despite the clear challenges in their relationship.
Trump and Xi had discussed the possibility of consultations on AI development during their meeting in Beijing in May, but no formal framework for those consultations was established. The new proposal comes as the two countries compete for leadership of the sector, alongside shared concerns that advanced systems could be used in cyber-attacks or activities affecting national security.
Trade talks move towards cutting tariffs on specific goods
Sunday’s talks also addressed a plan to activate a process agreed by the two presidents in May, allowing each side to identify non-strategic goods on which tariffs could be reduced. US officials called the process the “Trade Council”, and Greer said the two sides were working to turn the related ideas into workable measures.
Greer did not identify a final list of goods eligible for tariff reductions, but said the discussion focused on a relatively small group of non-sensitive US and Chinese products that could be traded in a balanced way. He indicated that potential Chinese exports could include consumer goods and low-tech products, while US exports could include energy products, agricultural goods and medical devices.
Greer said before the talks that the Trump administration would continue seeking more balanced trade with China by following up on previous commitments, improving the flow of non-sensitive goods, and expanding access to the Chinese market for US farmers, manufacturers and workers.
US demands for safeguards to prevent misuse of technology
Democratic Representative Ro Khanna, a senior member of the House Select Committee on the Strategic Competition between the United States and the Chinese Communist Party, said any understanding between Trump and Xi on AI safety should include a ban on systems capable of iterative self-improvement, as well as safeguards to prevent the misuse of the technology in developing biological or nuclear weapons.
The economic talks aim to pave the way for potential deals during the Washington summit, as the existing trade truce between the two countries approaches its expiry date on 10 November.
The truce eased the tariff confrontation between the world’s two largest economies, but left issues involving tariffs, technology, critical minerals and commercial purchases under negotiation.
