What You’re Getting Wrong About China And Ai

To hear San Francisco’s techies or Washington’s policy wonks tell it, the AI competition between the U.S. and China is the “Space Race” of our day. Only this time, the prize isn’t satellites orbiting the earth or footprints and flags frozen on the moon — it’s limitless economic and military potential.

But according to journalist Yi-Ling Liu, the author of the new book, The Wall Dancers: Searching for Freedom and Connection on the Chinese Internet, the narrative of an existential AI race between China and the U.S. risks becoming a self-fulfilling prophecy. And she questions whether the tech hawkishness coming from Silicon Valley interests and their ideological allies is more about the pursuit of their own deregulatory agenda, even if it means making AI and the world less safe.

“The biggest risk is this rolling back of regulation and accelerating ahead without safety parameters in the name of trying to beat [China],” she says in a new interview with POLITICO Magazine.

Liu also explains the wave of “China envy” sweeping Silicon Valley, why the Trump administration may be unintentionally encouraging other countries to be more reliant on Chinese tech and how the world’s two greatest superpowers could join together to avert an AI catastrophe.

This interview has been edited for length and clarity.

We often hear that the U.S. and China are engaged in an AI arms race. But you write that seeing it as a zero-sum game is counterproductive, and that the two countries need to work together to ensure emerging technologies are safe. That doesn’t seem to be happening any time soon. What are the biggest risks in this competition between the two great powers?

Maybe one of the biggest issues and problems with this narrative of the race is that it might not actually be describing reality. Especially after the launch of [Chinese open-source AI model] DeepSeek-R1 last year — it’s a so-called “Sputnik moment” that kind of threw U.S. policymakers and industry people into a frenzy. This idea that both countries were somehow racing towards AGI, or artificial general intelligence, when we in fact have very little evidence that AGI is a national priority for China. And so one of the biggest risks of furthering this particular narrative is making it a self-fulfilling prophecy, creating a story where it didn't exist in the first place — and not stepping back and thinking about who are the people who are creating this narrative? To what extent is this narrative created by industry players who are trying to achieve their own corporate goals?

From the U.S. side, there's this idea: “If we don't move as fast as possible, China is going to beat us.” There's going to be a huge attempt to cut down and cut back on regulation, from creating sufficient safety evaluations and benchmarks on frontier AI models and making sure that labs are held accountable to the frontier safety policies they're putting out — instead of emboldening them to be like, “Well, go and speed ahead and accelerate, it doesn't matter if you are stepping across a line that might make a model a potential biohazard or present potential bio risks, because that might be directly contradictory to trying to beat China.”

The same can be said for things like intellectual property. Right now in the U.S., there's a huge ideological debate on, to what extent can we feed these models the intellectual property of artists and writers? And there could be a rolling back of regulation in response to claims that China, for example, has really limited intellectual property rights, and they're just feeding their models everything. And so I think the biggest risk is this rolling back of regulation and accelerating ahead without safety parameters in the name of trying to beat the other.

Is there an alternative to the arms race? What would a more collaborative approach look like?

One of the smallest things that is happening right now that I think could be done at a much larger scale are track-two dialogues between the two countries. Track-one dialogues are dialogues that take place between government leaders — heads of state, presidents, prime ministers. Track-two dialogues are those that take place between scientists on both sides, civil society organizations on both sides, people who have direct levels of influence in the government on both sides.

It’s very important to bring these two parties together to discuss, for example, red lines that won't be crossed or established, like a common set of safety standards for models, to just get these people together and have some sense of dialogue and common ground. And to have a sense of what's going on in each country before jumping to conclusions and using those poorly sketched-out conclusions to drive policy back home.

You write about a wave of “China envy” sweeping policy circles and Silicon Valley. What is making them so jealous?

Yeah, this is definitely a dynamic of the past year. I can trace the roots to two explicit events: I would say the first is probably the launch of DeepSeek-R1. Everyone was freaking out about that.

I'd say the other, funnily enough, was the moment when a lot of TikTok refugees were migrating to RedNote, the Chinese social media app, in the wake of the TikTok ban. I would say that it really started more among policy wonks — the intellectual, chattering-class, abundance folks. I think this also spans both the left and the right, where China was becoming this rhetorical mirror onto which Americans started to project their fears and aspirations.

I started to notice a bunch of pundits and CEOs and Silicon Valley tech people all going to China for these two-week trips to chronicle their observations of Chinese technology. And it was almost always through the lens of American stagnation. There was always this, “Oh my gosh, China can build humanoids! China can build bridges!” There's this obsession with China's ability to build high-speed rail, despite the fact that Chinese people have been building high-speed rail for like a decade now. I think this kind of obsession with China's ability to build things and build infrastructure in the physical world is at the root of a lot of this envy, and a lot of it is rooted in a U.S. realization and awareness of their own inability to build infrastructure in the physical world.

I think it also is driven by this awareness of the erosion of the U.S.’ own political system. I think that's driving a lot of more of the China envy on the popular front, within the mainstream consciousness, where you see people like the influencer IShowSpeed, or Hasan Piker going to China and doing these little YouTube videos, talking about how good China is at building drones and buildings. It's much less about China itself. If I were to talk to people in China about how much their world has transformed in the past year? Not that radically. It's not like it suddenly became shinier and bigger and better. It's more what has shifted in Americans’ perspective of China.

There’s a long history of Silicon Valley tech CEOs having China envy. Back in 2019, when WeChat was becoming huge — I mean, it's always been this huge platform, but I remember Mark Zuckerberg being described as having “WeChat envy” because he wanted to build Facebook into the monolith that WeChat was, where it was not just a messaging platform, but had its tentacles in every facet of public life. I think Elon Musk definitely has some level of China envy, where he wants to be able to move fast and break things and build things, and there definitely is this sense of envy in Silicon Valley with the speed with which China moves.

As the U.S. pulls back from the world stage and allies seek “digital sovereignty,” how might that reshape other countries’ relationship with China?

That’s something I'm going to be watching closely. For a long time among a lot of U.S. allies, there's a sense that [the U.S.] is a reliable player that we can count on, this is a digital ecosystem that we can be part of. Even if there were rules around AI governance, for example, these are rules that we can co-create and assume that the United States will also participate in. And that's essentially thrown out of the window.

I think the question that a lot of other countries now face is, “In this increasingly or already multipolar world, how are we going to navigate these two gigantic superpowers? Where are we going to fit in? How much more dependent will we have to be on China when this relationship with the U.S. is no longer sustainable?”

I think this is going to mean much greater dependency on China from other countries with regards to hardware. Europe is going to have to think very deeply about its car market and EVs, and what that relationship is going to look like with China. The Middle East — I think everywhere is going to have to think about to what extent they're going to rely on digital infrastructure from China, humanoids from China. So much advanced technology is going to be coming out of China that navigating that dependency is going to be a crucial question for many countries in the years to come.

When it comes to AI governance as well, other countries are going to have to rely on China to be a more reliable player and stakeholder in drafting up these agreements, defining standards, etc., in a way that the U.S. used to play a much more central role.

The “dance” in your book refers in part to the push and pull on the Chinese internet between government crackdowns and censorship and public freedom of expression and connection. How has that dance impacted your view of how countries like the U.S. approach internet freedoms? Is America doing a “dance” of its own?

Absolutely. And this is something that I didn't expect as I started writing this book. One of the examples that I can just think off the top of my head is this recent TikTok sale. For a long time, we were thinking about TikTok as a dance between the Chinese state and Chinese tech companies and the Chinese people — or later on, the Chinese state, Chinese tech companies and the rest of the world, the American people. To what extent can Bytedance — given its control of the algorithm — or the Chinese state — given its relationship to Bytedance — shape what's appearing on our feeds? And to what extent can users not be used by that, or adapt to that or find creative ways around Chinese censorship domestically as well?

People were asking me about the implications of this TikTok sale. Sure, the owner has changed. The majority stake is going to be held by a set of American investors. But the question now is to what extent does Oracle have control over the content moderation of this platform and what appears on these feeds? To what extent does Oracle's relationship with the Trump administration shape what's taking place on these feeds? And then in response to that, how much power do users have to bypass that influence, or be shaped by or shape that influence? And so, what we once saw as a dance that was unique between the Chinese government, Chinese tech companies and the Chinese people, I feel like is now very much playing out on U.S. turf.

I think the other thing we're seeing that I didn't realize was just the relationship between Silicon Valley's tech leaders and the Trump administration. Back when I was reporting in China, it was obvious to me that there was this relationship of patronage and interdependency between the Chinese government and Chinese tech companies. Chinese tech companies provided innovation, but they needed the government stamp of approval to get policies across, or bypass regulations or to operate in the way they wanted to operate. And I was really struck to see last year, after the inauguration, essentially all of Silicon Valley's big tech CEOs showing up to kowtow to the new leader and align their rhetoric and the direction of their policies such that it kind of follows — you know, dances with, so to speak — the tune of the current administration.

What are some of the biggest misconceptions you encounter about China, the internet and technology, particularly in American politics?

China is always addressed as a monolith. For some reason, there is this feeling that everyone is the same. This has historically been the case and continues to be the case, that it serves as this kind of two-dimensional projection onto which people in the U.S. project their fears and desires. I think the other assumption that's tied to that is, the Chinese state is a monolith — and therefore, in relation to that, you can only be a patriot or a dissident, you can only be an apologist or an opponent of the state. A Chinese person can only have one of two identities, when, obviously, it is possible to hold a whole, wide spectrum of identities and relationships in relation to the state.

Those are the two main [misconceptions], and also that it's possible for a place to hold two seemingly contradictory identities at once. It is possible that China is a technological powerhouse and building some of the most advanced technologies in the world, but also economically stagnant, and its people are struggling with employment. They are feeling really down about future prospects. It's suffering from demographic decline.

For some reason, in the American public imagination, it's only possible for China to be this superpower economic juggernaut or this stunning oppressive regime where people have no agency. It’s very important for us to be able to hold the multitude of different identities that this one country can be and people can be.