The United States has proposed creating a bilateral AI notification mechanism with China — effectively a hotline that would allow each side to warn the other when an AI-related incident threatens national security. The idea was floated during Sunday talks in New York between Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng, ahead of Xi Jinping's state visit to Washington this week. No details on what would trigger a notification have been publicly revealed. The proposal sits within a broader US push for a dedicated US-China AI dialogue. Bessent framed it as a transparency initiative between "the number one and the number two AI powers in the world." Trump science adviser Michael Kratsios called the focus "avoiding shared risks." Carnegie Endowment's Scott Singer endorsed the concept as "a productive first step" and "an obvious win for both sides." Beijing, characteristically, said almost nothing — Xinhua confirmed AI was discussed but did not acknowledge the notification mechanism. The historical precedent is mixed. The US and China have had a direct secure telephone link since 1997 and a military hotline since 2008, but the track record inspires limited confidence. The military line gives both sides 48 hours to respond, and US officials have complained that sometimes no one picks up. After the US shot down a suspected Chinese surveillance balloon in 2023, then-Defense Secretary Lloyd Austin sought a call with his Chinese counterpart and Beijing declined. The urgency behind the proposal is real. Anthropic CEO Dario Amodei warned this month that within 6-12 months, an AI swarm could be "capable of taking over the entire internet with a persistent botnet," potentially causing hundreds of billions in damage. OpenAI's Sam Altman and Elon Musk endorsed the assessment. Trump, meanwhile, has resisted calls for an AI development slowdown, insisting "whoever wins AI, wins." The investment asymmetry is staggering. According to Stanford's 2026 AI Index Report, US companies invested $285.9 billion in AI in 2025 versus $12.4 billion in China — a 23:1 ratio. But China has been closing the gap through efficiency: DeepSeek's January release, built at a fraction of competitors' costs, sent US tech stocks plummeting. China also holds structural advantages in rare earth mineral production and electricity generation — more than double US capacity — providing cheap power for AI data centers. The AI hotline will compete for diplomatic bandwidth at a summit already overloaded with unresolved trade disputes, tariff enforcement, Russia sanctions affecting Chinese and Indian energy purchases, the US-Iran conflict where China is an economic ally of Tehran, Taiwan, and rare-earth supply dominance. Singer argues that previous AI dialogue in 2024 "failed because frontier AI risks got mixed in with other issues," and a dedicated channel could prevent that. Whether Beijing sees the same value in transparency that Washington does remains entirely unclear. The structural question is whether this mechanism creates any genuine constraint or merely provides diplomatic cover for an unrestricted AI arms race. A hotline without defined triggers, verification protocols, or enforcement teeth is an agreement to agree later. The concept is sound — nuclear hotlines demonstrably reduced miscalculation risk during the Cold War. But AI incidents are faster, harder to attribute, and less visible than missile launches. The mechanism's value will be determined entirely by details that do not yet exist.
