Trump Pushes AI Hotline With China Before Year's End

The Trump administration is moving fast on a proposal that could reshape how the US and China manage their most volatile technological competition: an AI notification system designed to alert each other about national security threats related to artificial intelligence.

During President Trump’s state dinner with Chinese President Xi Jinping in Washington this week, the initiative has taken center stage alongside tech heavyweights from OpenAI and Nvidia. Treasury Secretary Scott Bessent first unveiled the notification system on Sunday during talks with Chinese Vice Premier He Lifeng, signaling serious momentum on both sides.

A Cold War Parallel for the AI Age

The proposal essentially mirrors the famous hotline that connected Washington and Moscow during the Cold War, except this time the focus is keeping tabs on AI developments that could threaten either nation’s security. It’s a recognition that frontier AI models are advancing so rapidly that neither country wants to be caught off guard.

The framework would establish clear benchmarks for what constitutes a national security incident worthy of alerting the other side. Sounds straightforward, but the devil is in the details. Officials inside the Trump administration acknowledge this mechanism will take several more weeks to finalize, even though they’re optimistic about launching it before 2026 ends.

The real challenge? Getting China to actually commit. Beijing has been noncommittal so far, and enforcement measures remain thorny territory. Trump officials want to build in safeguards to ensure it’s not just the US constantly alerting China while receiving radio silence in return.

The Chip Export Complication

Here’s where things get messy. The Trump administration is planning to extend a trade truce with China while keeping restrictions on exporting cutting-edge Nvidia chips. But developing an AI notification system with real teeth requires enforcement mechanisms that need White House sign-off, and that approval could get tangled up in broader trade negotiations.

Nvidia has been granted limited licenses this year to sell its H200 chips to China, components essential for AI labs to scale up their computing power. Nvidia’s advocates argue that preventing these sales only pushes Chinese companies like Huawei to develop their own alternatives, though recent reports suggest Huawei has already cracked that problem with commercially available AI chips.

Who Gets the AI Czar Crown?

Another wrinkle complicating the timeline: the Trump administration still hasn’t decided who will serve as its “AI czar” to lead the new “AI Force.” This interagency coordinator role could influence how the notification system actually operates. Bessent’s name has circulated heavily, partly because of his parallel to Hank Paulson’s role managing US-China relations during the 2008 financial crisis. Paulson later credited that communication channel with preventing China from dumping massive amounts of US Treasury debt.

But Bessent isn’t the only contender. Commerce Secretary Howard Lutnick, Michael Kratsios, and Sean Cairncross are also in the mix. Unlike his predecessor David Sacks, who served as a special government employee, it remains unclear whether Trump’s choice will wear two hats or step down from their current position entirely.

Lutnick has supporters and critics within the administration. While some Trump officials have resisted giving him more AI authority, he’s stayed competitive partly because he oversees the Center for AI Standards and Innovation, which handled frontier AI crises competently this summer.

The Real Stakes

Bessent and Lifeng are scheduled to meet again in two months in Shenzhen, China, and that’s the Trump administration’s internal deadline to have a functional framework finalized and approved. The next steps matter enormously, not just for bilateral relations but for how the global AI race unfolds. The US and China remain locked in a sprint to dominate frontier AI capabilities, and a notification system could either ease tensions or create new friction points if either side feels the other is gaming the rules.

The talks underscore a fundamental shift: treating AI development like a Cold War arms race requires Cold War-era communication protocols. Whether that’s enough to prevent catastrophic miscalculation remains the trillion-dollar question.

This article is based on reporting from Hugo Lowell’s Inner Loop newsletter on WIRED.