When Trump and Xi meet they may focus on AI. ‘Observe Two’ talks are already buzzing : NPR
Chinese language and US flags wave exterior a expertise firm in Beijing, on April 17, 2025. (Photograph by PEDRO PARDO/AFP through Getty Photos)
AFP
conceal caption
toggle caption
AFP
The US and China are locked in a cycle of fierce competitors and distrust over synthetic intelligence, and consultants say the possibilities that President Trump and Chinese language chief Xi Jinping can change the dynamic once they meet in Washington subsequent Thursday are dim.
Previously few weeks alone, the Trump administration has accused Chinese language AI firms of stealing mental property by “distilling” main U.S. AI fashions, and threatened sanctions. Beijing, for its half, has poured scorn on U.S. restrictions on chip exports, and accused Washington of operating a “Chilly Warfare playbook” to carry Chinese language AI down.
It could appear there’s little room, or urge for food, for compromise. However former officers and coverage consultants say there’s a buzzing bilateral dialogue — and a rising sense of shared concern about the hazards of AI — on the sub-national degree in what diplomacy wonks name “Observe Two.”
For many years, government-to-government interactions, generally referred to as “Observe One” diplomacy, have been underpinned by this sprawling community of casual personal conferences and conferences, typically run by assume tanks or universities. Observe Two discussions carry collectively consultants from either side, and canopy a wide selection of matters, from nuclear disarmament to local weather change.
“I feel there isn’t a extra vital subject that the U.S. and China needs to be speaking about than AI,” stated Sarah Beran, a former U.S. diplomat who labored on China and is the incoming president of the non-profit Nationwide Committee on U.S.-China Relations. The committee runs a number of Observe Two discussions and has been concerned in coverage exchanges since 1972, when it sponsored a go to to China by U.S. ping pong gamers to assist thaw relations.
AI is a troublesome subject, although, as a result of there’s so little belief within the relationship in the intervening time, says Beran. “The worry of chokepoints, being held again, and AI getting used as a cudgel on either side signifies that that is additionally one of the troublesome areas to speak about,” she stated.
The concept behind Observe Two is to construct understanding, and work towards coverage suggestions that either side can agree on. Generally, governments use them as again channels, or low-stakes methods to drift concepts.
Beran stated that was the case after direct military-to-military discussions between China and the U.S. had been lower off following then-Speaker of the Home Nancy Pelosi’s go to to Taiwan in 2022. On the time, Beran was the highest China official on the Nationwide Safety Council within the Biden administration.
“We floated some concepts in Observe Two dialogues about what resumption of military-to-military communications might appear like, at what degree, what the matters can be. After which we took it again into the formal channels and negotiated on resuming these channels,” she stated.
Then-President Biden and Xi agreed at a 2023 summit in Woodside, Calif., to get the army dialogue again up and operating.
Jack Shanahan, a retired Air Power basic who has been concerned in Observe Two dialogue over army AI, stated casual discussions helped tee up an settlement between the leaders a 12 months later that any determination to make use of nuclear weapons needs to be managed by people, not by AI.
“That was solely doable by years of quiet dialogue on the Observe Two degree on this very subject,” he stated.
Samm Sacks, an skilled in Chinese language tech on the Johns Hopkins Faculty of Superior Worldwide Research and the coverage assume tank New America, stated she was concerned in a number of Observe Two discussions on AI in Shanghai and Beijing over the summer season “that had been very substantive.”
“Folks can discuss delicate points extra candidly and in addition do not feel the necessity to come to any consensus that might change into public,” she stated.
These dialogues have strengthened for her that the U.S. has plenty of widespread floor with China on AI, regardless of authorities friction over chip export controls and accusations that Chinese language AI labs are stealing mental property. “So what I hear within the Observe Twos I am part of is a way of shared vulnerability and shared precarity created by AI,” she stated.
Shanahan agrees. On AI, China sees itself as an underdog, he says, however Shanahan has perceived a shift on the Chinese language aspect because the expertise advances and dangers change into clearer. “For the primary time, I am seeing, like, ‘Look, we all know we’re not at parity in a few of these areas, however AI is simply too vital to not have these dialogues,'” he stated.
Nonetheless, tensions over AI on the official degree have been excessive, and the 2 sides have tended to deal with completely different considerations, with the U.S. taking a look at issues like cyber vulnerabilities and doable lack of management of AI, and China occupied with regime stability and self-reliance for making {hardware}. Specialists say that has an affect on the complexion of Observe Two dialogues — and has tempered expectations for concrete outcomes.
The Heart for Strategic and Worldwide Research, a U.S. coverage assume tank, additionally runs Observe Twos, and China specialist Scott Kennedy has been concerned in a wide-ranging dialogue about economics that features AI. He stated the talks can improve consciousness in coverage circles of the others’ purple traces and the way both sides makes selections on the problems.
“They supply you a kind of highway map of the completely different factors of view and the potential actions and reactions of various situations,” he stated. However he added they require a degree of presidency help on either side to work.
“That is the time the place we’d like some management, and what we’re seeing is a vacuum,” he stated. “Observe Twos do not fill that vacuum. They may also help redirect issues and purchase time, put together the groundwork for brand new engagements, however they don’t seem to be going to unravel the large issues on their very own.”
Sacks stated the disagreements over AI between China and the US make this a precarious second, with the utility of Observe Twos within the stability. A push this week by U.S. AI firms for a world settlement to decelerate the event of innovative AI didn’t go over properly in China.
“I really assume one of the best factor that the 2 governments can do is simply at a excessive degree sign [that] we each want to come back collectively to handle AI security and cyber threat, bio threat, for instance,” she stated, referring to the dangers that individuals might use AI for hacking or to develop organic weapons.
Some Observe Two contributors say the Trump administration is much less engaged with the dialogues than earlier administrations, making the work more difficult. However Observe Two contributors might be watching the Trump-Xi assembly in Washington subsequent week for alerts that AI is a subject of concern to either side, and one on which the 2 sides’ consultants can proceed to have interaction in substantive talks. And Sarah Beran says that features the Nationwide Committee.
“You should construct confidence on either side that it is a protected house to speak about points,” she stated. “Our subsequent assembly is arising within the subsequent couple of months and we’re making an attempt to consider how we use regardless of the two presidents say collectively as a approach to begin speaking about these points in a Observe Two format.”
