A Cold War‑style calculation is shaping this week’s Trump–Xi summit in Washington: U.S. and Chinese officials are considering a direct crisis communication channel — a so‑called "red phone" — to use if a severe artificial intelligence (AI) incident occurs.
Why it matters
U.S.–China ties are fragile enough that a dedicated AI crisis line could be the most tangible outcome of the summit, potentially more significant than trade deals, a slowdown in strategic competition, or a diplomatic reset.
Current posture
President Donald Trump has spent the past week attacking what he calls AI "doomers" and framing the competition with China as a single imperative: beat Beijing. He has labeled fears of an AI catastrophe a "HOAX" and vowed the U.S. will ‘‘not in any way hinder or stifle’’ AI while it seeks to keep its edge over China.
Leading American AI labs largely share the urgency about China’s rise, but many reject Trump’s dismissal of safety concerns: they argue that the intensity of the race is precisely why international safeguards are necessary.
China’s perspective
China is equally determined to lead in AI. Xi Jinping arrives in Washington on Wednesday for his first state visit since 2015 and is balancing two competing concerns:
- Falling behind: Xi does not want a safety regime that effectively freezes the race while the United States remains ahead. Beijing is promoting inexpensive, capable open‑source models globally and characterizes U.S. chip controls and calls for restraint as deliberate attempts to contain China.
- Losing control: Xi wants AI powerful enough to transform China’s economy but not so powerful that it slips beyond the Chinese Communist Party’s control. Beijing worries less about Silicon Valley‑style extinction scenarios and more about nearer‑term threats such as cyberattacks, malicious use, and disruptions to critical infrastructure.
Specific proposal and precedent
That narrow overlap of interests is why U.S. Treasury Secretary Scott Bessent pitched Chinese officials over the weekend on a "notification mechanism" for AI incidents with national‑security implications. The Cold War precedent is the hotline Washington and Moscow set up after the Cuban Missile Crisis — a direct channel intended to prevent accidents and ambiguity from spiraling into nuclear confrontation.
Trump and Xi are expected to discuss shared AI risks during their Thursday meeting, and officials are exploring a formal U.S.–China AI dialogue that could give a crisis channel a more permanent institutional home.
Practical challenges
To date, no party has publicly defined what would trigger the red phone. In a major AI disaster — for example, if a frontier model launched a widespread cyberattack against critical infrastructure — the first alarm might come from inside a private company rather than a military command center.
That raises thorny questions: the lab operating the model may know more about its behavior than government agencies, while officials would need to decide how much sensitive information to share before they fully understand the threat.
Patterns of breakdown
Washington and Beijing already maintain crisis hotlines, yet they have a history of going silent when channels are needed most. China suspended military contacts after Nancy Pelosi’s 2022 visit to Taiwan. Months later, Beijing declined a Pentagon request for a call after the U.S. shot down a Chinese surveillance balloon — precisely the sort of tense, ambiguous moments that hotlines were meant to defuse.
The COVID‑19 pandemic supplied another warning: a cross‑border threat that required transparency and cooperation instead spawned a prolonged U.S.–China blame game over origins, information sharing, and responsibility.
Bottom line
The mere proposal of an AI crisis hotline signals that AI is no longer just another theater of competition between the United States and China. A serious AI failure on either side could rapidly become a shared crisis, which gives both governments an incentive to consider a direct, reliable communications channel — even though the concrete rules for triggering it, the role of private firms, and the scope of information to be shared remain publicly undefined.
