U.S. Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng agreed in September to establish a bilateral dialogue on AI and a communication channel for AI safety incidents, setting the stage for technical discussions ahead of a meeting between Presidents Xi and Trump.
The diplomatic push follows mounting competition between the two leading artificial intelligence powers. During a September meeting at JPMorgan Chase’s New York City offices in the lead-up to Xi’s White House visit, U.S. officials proposed the notification mechanism to handle AI-related national security incidents.
Officials Establish Crisis Communication Channels
Bessent framed the diplomatic effort around transparency between the world’s top two AI superpowers, stating: We think that just like with any cross-border activity, that moving from opaque to more transparency between the No. 1 and the No. 2 AI powers in the world is very important.
The bilateral framework aims to provide a forum for exchanging views on cutting-edge AI models while establishing a direct communication channel for critical incidents. This notification mechanism could address severe technical anomalies, runaway models, rogue agents, or major model failures capable of triggering national security crises. U.S. export controls on advanced semiconductors and manufacturing equipment remain excluded from these safety talks. Subsequent technical sessions are planned for November 2026, and will likely take place prior to the upcoming summit between Presidents Xi and Trump scheduled for November 18-19 in Shenzhen, China. On November 23, the Center for Technology Innovation and John L. Thornton China Center at Brookings will host a webinar to explore how the United States and China are advancing the official bilateral China-U.S. AI Dialogue alongside the emergency incident communication link.
Draws Historical Parallels to Cold War Confidence-Building Measures
The dialogue draws historical parallels to Cold War confidence-building measures, such as the 1963 U.S.-Soviet bilateral hotline established after the Cuban Missile Crisis, though experts emphasize that these frameworks lack traditional arms control verification mechanisms. A further Cold War transparency and confidence agreement is the 1972 U.S.-Soviet Incidents at Sea Agreement, which was designed to mitigate maritime crisis hazards.
China’s 2021 official position paper on military AI applications explicitly acknowledged the risks of the deepening of strategic miscalculation, undermining of strategic mutual trust, escalation of conflicts, and damaging global strategic balance and stability.
Relies on Proprietary Closed-Model Ecosystems Protected by Initiatives
President Donald Trump declared in September, We’re leading China in AI […] and, frankly, I want to keep it that way because whoever wins AI wins,
making the lead in frontier models the measure of American ambition. The United States relies on proprietary, closed-model ecosystems protected by initiatives like the late-2025 Pax Silica pact—which counts more than 20 trusted partners including the EU and India—to secure supply chains and critical minerals. America consumer spending is deep. An L5 engineer in the Bay Area can pull in a total compensation package of $300,000, translating to more than $150 for every effective hour worked.
In contrast, Beijing promotes open-weight models through initiatives such as the World AI Cooperation Organization. Following the inauguration of the World AI Cooperation Organization, Beijing touts its openness and inclusiveness.
Nations Caught Between Competing Standards
Third-party nations with vital resources face pressure regarding both blocs. Local officials report that Chinese foreign direct investment in Kazakhstan reached a record $2.8 billion in 2025.
We don’t want this binary choice,
Masalimova said. Similarly, Mari Pangestu, vice chair of the Indonesian National Economic Council, stated that Indonesia intends to engage with both superpowers to select the best systems available. Like Kazakhstan, Indonesia has resources vital for AI and other technologies, such as nickel and copper.