U.S.-China 'Track Two' AI Talks Gain Momentum Before Summit

As President Trump prepares to meet Chinese leader Xi Jinping in Washington next week, the official relationship between the two countries over artificial intelligence remains fraught with accusations and mistrust. Yet beneath the surface, a network of informal diplomatic channels is quietly building common ground on AI risks and safety.

These "Track Two" dialogues—private meetings convened by think tanks, universities, and nonprofits—bring together former officials, policy experts, and technical specialists from both nations to discuss sensitive topics away from the glare of official negotiations. According to NPR, which first reported these details, AI has become one of the most active areas of Track Two engagement between the United States and China.

Why it matters

With AI development accelerating and both nations viewing the technology as strategically critical, the risk of miscalculation or conflict is real. Track Two channels provide a testing ground for policy ideas and help each side understand the other's red lines—work that can lay groundwork for official agreements even when government-to-government relations are strained. The upcoming summit will signal whether leaders are willing to translate this informal progress into concrete cooperation.

Building trust through informal channels

Sarah Beran, incoming president of the National Committee on U.S.-China Relations and a former National Security Council official, told NPR that AI represents both the most important and most difficult topic for bilateral discussion. The lack of trust makes conversations challenging, but Track Two forums offer a space where participants can explore sensitive issues candidly without needing to reach public consensus.

Samm Sacks, a Chinese tech expert at Johns Hopkins School of Advanced International Studies, participated in several Track Two discussions in Shanghai and Beijing over the summer. She described them as "very substantive" and said they revealed "a sense of shared vulnerability and shared precarity created by AI" on both sides.

Retired Air Force General Jack Shanahan, who has participated in Track Two dialogues on military AI, noted that years of quiet discussion at this level helped produce the 2024 agreement between Biden and Xi that humans, not AI systems, should control nuclear weapons decisions.

Different priorities, common concerns

The two countries have historically focused on different AI risks. U.S. concerns center on cyber vulnerabilities and potential loss of control over advanced systems, while China emphasizes regime stability and hardware self-reliance. Despite these differences, Track Two participants report a shift in Chinese attitudes as AI capabilities advance and risks become clearer.

"For the first time, I'm seeing, like, 'Look, we know we're not at parity in some of these areas, but AI is too important not to have these dialogues,'" Shanahan told NPR.

Scott Kennedy of the Center for Strategic and International Studies, which runs Track Two dialogues on economics including AI, emphasized that these discussions can map out different viewpoints and potential reactions to various scenarios. However, he cautioned that they require government support on both sides to be effective and cannot solve major problems on their own.

Limits and opportunities

Several Track Two participants noted that the Trump administration appears less engaged with these dialogues than previous administrations, complicating the work. Recent tensions—including U.S. accusations that Chinese AI companies are stealing intellectual property through model "distilling" and Chinese criticism of U.S. chip export controls—have created a precarious moment.

Beran said her organization's next Track Two meeting will occur in the coming months, with participants hoping to build on whatever signals emerge from the Trump-Xi summit. "You need to build confidence on both sides that this is a safe space to talk about these issues," she said.

These details were first reported by NPR.