-
Click here to listen to this article - Share via
See more from the L.A. Times in Google Search. Set us as preferred
As President Trump welcomes Chinese President Xi Jinping to the White House on Thursday for a state visit awash in pomp and pageantry, Washington is in the throes of full-scale panic over a potential artificial intelligence apocalypse and what, if anything, to do about it. As former White House chief of staff Rahm Emanuel once famously said, one should never let a good crisis go to waste.
Given the U.S. and China’s mutual distrust and full-on AI race, the widespread skepticism that they could strike a meaningful deal to limit AI risk is well-founded. But ironically, Trump and Xi might be the only leaders in decades with the clout and latitude to pull off the impossible.
Will it happen? Unlikely. But the same was said about major arms control deals of the past, until it wasn’t.
The unusually leader-driven nature of current U.S.-China diplomacy, with Trump especially eager to play up his personal relationship with Xi, makes a top-down deal to place guardrails on frontier AI more plausible. Both leaders also have extraordinary control over politics at home, freeing their hands to make potentially unpopular moves.
Just imagine former President Biden, across from Xi in the Oval Office, agreeing to handcuff the biggest U.S. tech companies right before feting the Chinese Communist Party leader at a lavish state dinner. The political costs would be catastrophic, probably dogging his presidency for months.
Yet Trump, with his demonstrated immunity to the laws of political gravity, enjoys maximum flexibility to absorb flak and make concessions that China hawks and Silicon Valley titans would resist. The “Nixon goes to China” cliches bandied about these days are overused but possess a grain of truth.
It’s true that Trump, in just the last week, has rejected the idea of doing much of anything to rein in AI. But the president reverses himself all the time, usually without consequences. He’s done it before on AI, agreeing last year to sell Nvidia’s advanced H200 chips to China while brushing off concerns it would hand China an AI advantage.
Trump, as he’s wont to remind us, is a dealmaker at heart. A deal with Xi could even win Trump his long-coveted Nobel Peace Prize, as OpenAI Chief Executive Sam Altman recently suggested.
Trump also has history to fret about. He’s publicly spoken about his fears of being compared to President Hoover if the Iran war triggers an economic collapse. Presiding over an existential AI catastrophe would leave quite a legacy for whatever part of humanity remains.
It’s also true that China’s dominant storyline, echoed in state media, is that any talk about a slowdown is just an effort to constrain China, a misgiving that plays into the country’s broader historical narrative about the West.
Yet the political climate on AI is fast-changing on both sides of the Pacific.
In the U.S., the conversation is moving past the reflexive argument that losing the AI race to China is scarier than the specter of runaway AI. Congress, still controlled by Republicans for at least the next few months, seems to be waiting for a permission structure to start legislating. And the remarkable newfound consensus from Altman, Elon Musk and Dario Amodei creates a moment of opportunity we may never see again.
Like Trump, Xi faces little chance of being restrained by his own system. Now serving an unprecedented third term, Xi has dramatically consolidated power, weakening collective leadership and eliminating term limits. He’s even enshrined “Xi Jinping Thought,” a doctrine that emphasizes the Communist Party’s dominant role in increasing the nation’s global power, in China’s constitution. In today’s China, whatever Xi decides on a given topic carries the weight of state ideology and is unlikely to be questioned.
That said, Xi might be the harder sell. Unlike America’s notorious dealmaker in chief, Xi’s style of diplomacy tends to eschew compromise. He places far less of a premium on personal relationships than Trump does, and Xi has reason to suspect the U.S. might abandon an agreement if the president decides it no longer serves his interests.
At the same time, there are real reasons Xi may not be able to avoid engaging the U.S. on AI much longer.
In China, although there’s less urgency around the safety issues surrounding AI, concerns about it being used to stoke dissent or destabilize the regime are more salient. The revelation this month that a California firm used AI to quickly build a tool that could hack WeChat, the ubiquitous Chinese app used by more than 1 billion people, was the latest cause for alarm. Even Xi now speaks publicly about the risk of losing control over AI.
How could Trump and Xi paper over their countries’ mutual skepticism and distrust? Like any delicate diplomatic negotiation, the key is creating space for dual narratives — letting each leader spin the same deal to their own audience completely differently.
Trump could sell a deal through an America First lens: as a concession he extracted from China. He could frame it as locking in America’s current lead in the race and cracking down on Chinese state-backed firms that siphon American innovation through “distillation” of cutting-edge models.
Xi could make the opposite argument: that he forced the U.S. to adopt China’s preferred framework for global AI governance, cooperation and human control of AI systems. It would not go unnoticed in Beijing that any slowdown would buy China time to narrow the massive U.S. lead in advanced chips and processing power needed to train AI.
The more ambitious the deal, the more formidable the obstacles. An agreement to “pace the frontier,” or slow down the development of ever-more-powerful AI models, would require verifying compliance, and neither the U.S. nor China wants to let the other poke around in its most sensitive systems.
On the flip side, a vague agreement to keep talking or pursue constructive engagement would merely play into China’s strategy of stalling for time and portraying itself as the responsible actor without committing to much at all. Case in point: Although the U.S. and China announced a new “AI dialogue” this week in the run-up to Xi’s visit, Treasury Secretary Scott Bessent said the next meeting would be in two months, a cadence that seems seriously mismatched to the pace at which AI concerns are proliferating.
At a minimum, there are narrower, less threatening steps neither country should find hard to accept. Establishing an AI crisis “hotline” or emergency communications mechanism between Washington and Beijing could reduce the risk of an unintended lurch up the escalation ladder. Safeguards against non-state actors using AI to build bioweapons would benefit everyone.
The gravity of the moment suggests the world’s two most powerful leaders should go big. The odds are long. But the convergence of powerful leaders and urgent circumstances presents a rare opportunity Thursday in the Oval Office.
The moment may not come again.
Josh Lederman is a former NBC News foreign correspondent and Associated Press White House reporter who has reported repeatedly from China.
Insights
L.A. Times Insights delivers AI-generated analysis on Voices content to offer all points of view. Insights does not appear on any news articles.
Viewpoint
Perspectives
The following AI-generated content is powered by Perplexity. The Los Angeles Times editorial staff does not create or edit the content.
Ideas expressed in the piece
The article argues that President Trump and President Xi Jinping, because of their unusually concentrated political power and leader-driven diplomacy, may have greater freedom than previous U.S. and Chinese leaders to negotiate politically costly limits on frontier artificial intelligence.
The piece contends that Trump’s willingness to reverse positions, his self-image as a dealmaker and his interest in a historic legacy could make him unusually receptive to an agreement, even after recently opposing significant AI restrictions. A summit-level arrangement would also give both leaders an opportunity to present concessions as victories for their own countries.
The article suggests that Xi may have incentives to engage despite China’s preference for rapid AI development. Beijing is increasingly concerned about AI-enabled cyberattacks, social instability and the possibility of losing control over powerful systems; recent reporting likewise identifies cyber, biological and loss-of-control risks as potential areas of shared concern.[4][12]
The piece proposes that a workable agreement would need to accommodate incompatible domestic narratives: Trump could portray it as protecting America’s technological lead and stopping Chinese exploitation of U.S. innovation, while Xi could present it as recognition of China’s preferred emphasis on international cooperation and human control.
Rather than insisting immediately on a comprehensive treaty, the article identifies narrower measures as plausible first steps, including an emergency communications channel, safeguards against AI-assisted bioweapons development and cooperation on serious cyber incidents. Recent U.S.-China discussions have in fact included a proposed notification mechanism for AI incidents posing national-security risks.[8][10]
Different views on the topic
A central opposing view is that personal authority cannot substitute for institutions capable of defining, testing and enforcing AI commitments. Analysts say both governments lack a clear institutional structure combining technical model evaluation, national-security authority and diplomatic coordination, making a durable agreement difficult to implement.[1]
Skeptics further argue that frontier-AI commitments cannot presently be verified in the way traditional arms-control agreements were. There is no established technical method for independently confirming another country’s model capabilities, safety testing or compliance, and neither government is inclined to trust the other’s assurances.[9] Former officials have consequently characterized the basic problem as a lack of both trust and verifiability.[7]
The article’s expectation that the two leaders might slow development conflicts with security officials and lawmakers who view an AI pause as a strategic risk. Republican lawmakers have warned that delaying U.S. development could allow China to overtake American capabilities, while congressional debate has been slowed by fears of conceding technological ground.[1][3]
Another perspective holds that Washington should strengthen, rather than relax, export controls and technology protections. Analysts argue that Chinese firms’ alleged model distillation, remote access to American computing and military applications make safeguards against technology leakage a prerequisite for cooperation, not a concession to be traded away.[2][6]
Critics also question whether a broad “pacing” agreement could survive the countries’ conflicting strategic incentives. China views calls to slow AI as possible containment, while the United States doubts that Beijing would disclose violations or accept intrusive inspections; experts therefore favor limited information-sharing, incident notifications and crisis communications over an ambitious bilateral accord.[5][11]
Even proponents of engagement generally see the immediate objective as establishing a durable channel for discussion—not negotiating comprehensive guardrails at the summit. The emerging U.S.-China dialogue is expected to focus first on communication about shared risks, with more sensitive issues such as weaponization, infrastructure protection and cyberattacks left for later talks.[13][14]
More From the Los Angeles Times
Podcasts
Greenline Housing Foundation has become something of a “first aid kit” for Eaton fire survivors, offering financial services that range from temporary housing grants to land banking that have helped some families fund their paths back home.
California once bet big on offshore wind to help power its clean-energy future, but the Trump administration is now paying developers billions to walk away from their projects. As California’s remaining offshore-wind leases hang in the balance, the state is fighting back in court.
The Hillside Strangler murders terrorized Los Angeles in the late 1970s. The killers were a pair of sadistic cousins. One confessed. The other was set to stand trial—until L.A. prosecutors flinched.
