US-China AI Hotline Won't Be Ready For Weeks, Trump Officials Say

A proposed US-China AI hotline — modeled on the Cold War-era Moscow hotline — will take weeks to finalize, even as President Donald Trump and Chinese President Xi Jinping meet Thursday for a state dinner at the White House with AI at the top of the agenda.

Trump administration officials expect the AI notification system, announced Sunday by Treasury Secretary Scott Bessent alongside Chinese Vice Premier He Lifeng, to still require a number of weeks of work before an agreement is complete. Officials remain optimistic the mechanism could be in place before the end of the year, according to administration sources who spoke with Inner Loop.

The stakes are considerable. The system is designed to let each country alert the other about national security incidents related to AI, at a moment when Washington and Beijing are racing to dominate frontier AI development. Executives from OpenAI and Nvidia have confirmed they will attend Thursday's dinner, where the proposal sits among the key agenda items.

What still needs to be settled

Several parts of the framework remain unresolved. Chief among them: defining benchmarks for what kind of AI incident rises to the level of a national security issue — a threshold that remains unclear — and securing explicit sign-on from China, which has so far been noncommittal.

Trump officials say they are targeting a framework that is functional, and crucially agreed to by Trump himself, by the time Bessent meets He Lifeng again in follow-up talks in roughly two months in Shenzhen, China.

The administration has internally discussed laying out criteria for when each country is obligated to alert the other, along with enforcement measures to ensure the system does not become a one-way channel with only the US alerting the Chinese.

That enforcement question may prove the hardest part. With the US expected to extend a trade truce, and Trump reluctant to halt exports of cutting-edge Nvidia chips to China, building a framework with real teeth could take time to secure White House approval, sources say.

The Nvidia factor

The administration has granted a limited number of licenses this year allowing Nvidia to sell its H200 chip to China — a critical component for AI labs seeking more compute power to train their models. Nvidia's supporters have argued that blocking sales only incentivizes Chinese companies like Huawei to develop their own chip technology. According to recent reports, Huawei has already crossed that threshold with commercially available AI chips.

That dynamic complicates any enforcement regime tied to chip exports, one of the few levers the US holds in the AI competition.

The 'AI czar' vacuum

Some Trump officials also suggest the notification system could be delayed while Trump decides who to appoint as his next "AI czar" — a role envisioned as leading a so-called "AI Force" in charge of interagency coordination.

Bessent is one of the names floated for the job, though it is not clear he is the standout candidate. Commerce Secretary Howard Lutnick and White House AI officials Michael Kratsios and Sean Cairncross are also in the mix, two sources tell Inner Loop. It remains unclear whether Trump will pick someone who wears two hats or departs their current role entirely.

The previous AI czar, David Sacks, held the position as a special government employee. Sacks is not considered a candidate to return, the sources say.

The push for Bessent stems partly from history: the AI notification system is seen as akin to the US-China dialogue between former Treasury Secretary Hank Paulson and former Chinese Vice Premier of Finance Wang Qishan. During the 2008 financial crisis, Paulson credited that hotline with helping stave off China from dumping a large portion of US government-sponsored debt issued by Freddie Mac and Fannie Mae.

Lutnick remains in contention despite resistance from some Trump officials to giving him more AI power, in part because he oversees the Center for AI Standards and Innovation, which successfully navigated crises with frontier AI model companies this summer.

Why it matters

The negotiations over the hotline reflect a broader shift: both governments now treat frontier AI capabilities as a national security matter requiring dedicated crisis communication channels, not just a commercial rivalry. The outcome in Shenzhen in two months — a functioning framework, a signed agreement, or another delay — will signal whether the two countries can operationalize AI risk management while competing to dominate the technology itself.