Ahead of President Donald Trump’s meeting with Xi Jinping on Thursday, administration officials privately tell Inner Loop they see little room for progress in alleviating AI safety and espionage concerns.
As he departed the White House, Trump left it an open question whether he would press Xi on Chinese interference in AI or the Nvidia scandal, as part of what officials say is an effort to not antagonize China during the summit.
But the fallout from former Nvidia CEO Jensen Huang’s affair with Alice Huang, an executive at Megaspeed and suspected Chinese intelligence asset, has made any negotiations around AI development fraught for the world’s two largest economies, officials say.
Back in September, the Trump administration moved to set up an AI incident hotline between Treasury Secretary Scott Bessent and Chinese vice premier He Lifeng to account for situations just like this one. Unfortunately, the communication channel remains at the concept stage and is not operational.
Secretary of State Marco Rubio told reporters last week the US and Chinese governments had engaged in AI safety talks at a “technical level” since the hotline announcement, but admitted it was “still a work in progress.”
Trump officials have previously discussed putting together a framework that lays out criteria for when each country is obligated to alert the other and enforcement measures to ensure it doesn’t only become the US alerting the Chinese.
Part of the reason progress hasn’t been made yet is due to differing views on what the biggest risks are with AI—and both nations have different incentives on making incidents public, officials say.
In the US, Trump administration officials and tech leaders have been most worried about AI models becoming autonomous and misaligned, where they can operate beyond the bounds of human oversight.
But in China, government officials are widely understood to be more concerned about advanced open-weight models being used by critics and dissidents inside the country to undermine the Chinese Communist Party, officials say.
In an added complication, the Chinese government has appeared interested in using frontier AI models to gather intelligence domestically and abroad, and would be unlikely to report situations where their models went rogue, officials say.
Suspicions about mutual espionage have also exacerbated the distrust on any AI-related negotiations, officials tell Inner Loop. China has viewed the US as trying to stymie its AI development—and that was even before the Nvidia scandal.
China has forcefully bristled at allegations that Alice Huang was recruited by the Chinese intelligence agency, calling them “groundless” and a “smear.”
While Trump has sought to forge friendlier economic relations with China, the administration this week announced it would move forward with a new rule clamping down on China’s ability to remotely access US-made AI chips in foreign data centers.
It marked a direct intervention by Trump in response to the alleged role of Megaspeed in the Nvidia scandal and criticism that a firm with ties to China had gained access to advanced AI chips. The rule could also prevent similar companies that might have their corporate parents in China from doing business with Nvidia, WIRED has previously reported.
This is an edition of Hugo Lowell’s Inner Loop newsletter. Read previous newsletters here.






