The Missing Resistance in China’s AI Debate

Trump’s visit to China in May was marked by pomp and circumstance but produced few breakthroughs. One of the summit’s few concrete results was an agreement to cooperate on “artificial intelligence (AI) guardrails.” The United States and China are locked in a modern-day space race over AI, with both governments viewing AI leadership as central to economic competitiveness, military power, and global influence. At the same time, the potentially catastrophic risks posed by advanced AI—from military escalation and cyberattacks to loss of human control over powerful systems—make some degree of cooperation not only desirable but necessary.  But it remains unclear how the two countries would work together, given deep mistrust, fear on both sides that slowing technological development could mean losing out to the other, and differences in how the two sides define AI-related threats.

One area where China has a clear competitive edge is that, compared with the United States, AI appears to face much less popular resistance in the country. In a 47-country KPMG survey, 69 percent of respondents in China said AI’s benefits outweighed its risks, compared with only 35 percent who felt the same in the United States. Many Chinese users seem pragmatic about AI: If a tool is useful and affordable, they readily use it. Meanwhile, the leaders of Chinese AI companies quietly focus on building AI tools, not philosophizing about societal transformation or advising how Beijing should address it.  

China can look more practical and focused toward technological innovation vis a vis the United States. But beneath the order is a state that suppresses dissent and a society in which people have little say over the kind of technological future being built in their name.

A key difference between AI governance debates in the United States and China is the role of actors outside the state. In the United States, AI policy is shaped not only by the federal government but also by courts, technology companies, researchers, journalists, civil society organizations, and local communities. These actors often disagree with one another and bring different interests into the policy process. Their participation creates a multidirectional debate over how AI should be regulated.

In China, the process is centralized. AI governance is shaped primarily by state mandates and regulatory agencies. Within politically permitted boundaries, experts may offer advice and citizens may voice concerns online. These inputs, however, do not form independent centers of pressure capable of shaping AI governance outside the state-led framework.

As a result, in any future AI negotiations between the United States and China, Washington will enter the talks carrying the pressures of a contested democratic debate, while Beijing will enter with a more unified state position.

AI in China should be understood first and foremost as a state-led initiative. China’s AI ambitions have been built into the highest levels of state planning for years. In 2017, the government’s “New Generation Artificial Intelligence Development Plan” set the goal of making China the world’s primary AI innovation center by 2030. Since then, AI has been folded into Beijing’s broader strategy for technological self-reliance, industrial upgrading, and national security. The state treats AI not just as consumer technology or private-sector innovation, but as a strategic capability central to China’s economic modernization and geopolitical competition.

That priority has only intensified. In 2025, Chinese President Xi Jinping urged the government to mobilize national resources to accelerate the application of AI across the economy. This reflects China’s broader governance model: Once a technology is identified as strategically important, central and local governments, private companies, banks, and research institutions are pushed to move in the same direction, through policy directives, regulations, subsidies, state-backed financing, government procurement, and research funding.

Chinese public acceptance of AI should not be taken at face value. AI development in China is not unfolding in an open information environment. The Chinese government stringently censors information that it deems politically sensitive. Narratives that run contrary to state objectives rarely circulate freely. The state has already censored articles that urge the government to allow foreign AI tools, question whether China is becoming an “AI island” disconnected from the rest of the world, and warn that widespread use of AI in creative industries will drive down the quality of work.

One of the few prominent voices to challenge AI-enabled surveillance in China was Tsinghua University law professor Lao Dongyan. In 2019, she criticized the use of facial recognition in Beijing’s subway system. More recently, Lao warned that big data surveillance and a proposed national cyber identification system would greatly expand state control over citizens’ lives. Authorities responded by swiftly deleting her posts and suspending her social media account. She was also subjected to attacks in state media. Such criticism has become increasingly rare precisely because of reprisals such as those directed at Lao.

In November 2023, during a Halloween costume parade in Shanghai, one man dressed as a surveillance camera—an unmistakable symbol of China’s state surveillance system—was seen being questioned by police. The public never learned what happened to him afterward. Other participants also wore “provocative” costumes mocking state policies. Witnesses reported seeing authorities breaking up crowds and arresting revelers. The following year, Shanghai authorities restricted Halloween festivities to designated venues and, ahead of the event, issued warnings to participants. If protest against the surveillance state—a defining feature of contemporary China for the past decade—can be expressed only through costumes, and even then be suppressed, one can imagine how difficult it would be for anti-AI sentiment to emerge openly.

To be sure, not all criticism of AI is suppressed. Concerns such as about job displacement and scams can appear in Chinese public discourse, especially when it is framed as responsible governance. A Chinese court recently ruled that it was illegal for a company to fire an employee on the grounds that AI had taken his job. The case was widely discussed as an example of balancing innovation with labor protection.

And AI is only strengthening China’s sophisticated information control system that makes questioning AI difficult in the first place. AI can help platforms detect sensitive words, images, and coded language more effectively and identify viral posts before they spread widely. AI can also help state propaganda organs flood the information space with favored narratives, tailor messages to different audiences, and respond quickly to breaking events.

On generative AI, Chinese law requires providers and users to uphold “core socialist values” and prevent the production of content that the authorities say threatens national security or social stability. Testing of major Chinese large language models has found systematic limits on politically sensitive subjects such as Tiananmen, Xinjiang, Taiwan, and Xi Jinping. The extent of control moves beyond censoring particular facts to shaping the contours of permitted knowledge, delineating what is framed as controversial and what is presented as settled truth. For Beijing, ideological control and regime security are core to what constitutes “AI safety,” something Washington should keep in mind when negotiating with China on AI cooperation.

Also largely absent from China’s AI discourse is pushback from civil society. In the United States, digital rights organizations are warning that AI worsens surveillance, discrimination, privacy violations, and algorithmic manipulation. Labor groups are pressing companies and governments to address job displacement, workplace monitoring, and the use of AI to weaken workers’ bargaining power. Environmental activists are challenging the energy and water demands of data centers. Parents and children’s safety advocates are raising alarms about chatbots that can form emotionally manipulative relationships with minors. Lawsuits have begun to turn these concerns into legal claims, including cases brought by parents who allege that AI chatbots contributed to their children’s suicides. And some anti-AI activism have even taken a violent turn, prompting law enforcement warnings about “anti-tech extremism.”

That kind of pushback is largely absent in China because the organizations and mechanisms that might turn private concern into public pressure have been systematically dismantled. Since Xi Jinping came to power in 2012, the Chinese government has steadily crushed independent nongovernmental organizations, imprisoned human rights lawyers, and silenced free speech and labor rights activists, making policy advocacy outside state-sanctioned channels increasingly risky.

As a result, users may complain on the Chinese internet about fraud, bad products, or job losses, but there is no independent ecosystem of digital rights groups, labor unions, environmental organizations, parent advocates, investigative journalists, and public-interest lawyers that can openly challenge the state and major technology companies together. In China, people most affected by AI have few channels to organize and debate about what kind of AI future they want.

AI-enabled surveillance creates a self-reinforcing problem. The more the state deploys facial recognition and predictive policing, the harder it becomes for citizens to organize against those very tools. Activism requires people to find one another, communicate, and act collectively. AI surveillance makes each of those steps more visible to the state.

Granted, the party-state does not entirely ignore social concerns, but it addresses them from the top down, on terms set by the state. Beijing can move quickly to restrict AI products it sees as socially harmful, such as chatbots designed to create emotional dependency or addictions. But this is not the same as permitting society to become an autonomous source of political pressure.

The contrast between the United States and China in the AI discourse is also visible in the behavior of technology leaders. In the United States, AI executives increasingly speak as if they are not merely building products but designing civilization. Elon Musk has predicted a future in which AI and robotics make work optional, produce “universal high income,” and render money itself irrelevant. OpenAI has issued sweeping proposals for industrial policy in the “intelligence age.” Anthropic has laid out scenarios for global AI leadership, warning that democracies must stay ahead of authoritarian regimes.

Chinese AI companies operate in a different political universe than their counterparts in the United States. Their leaders understand that the largest questions—what AI is for, whose values it serves, what limits it must obey—are ultimately decided by the state. In contrast to American tech entrepreneurs such as Sam Altman and Elon Musk, Chinese tech founders do not present themselves as larger than their work. Liang Wenfeng, the CEO of DeepSeek, rarely grants interviews. When asked why he builds humanoids, Wang Xingxing, the chief executive of the humanoid startup Unitree, gives the same answer in every interview: It makes money. As tech analyst Afra Wang put it, “AI in China is not seen as an elite technology to be contained, nor as an anti-egalitarian threat. It is seen as the state’s instrument of Darwinian upgrade—and the instrument does not get to interpret itself. That work belongs to whoever holds the mandate. The companies build. The state then decides what it has been built for.”

This “I know my place” mentality among Chinese tech founders comes from learned experience. Chinese entrepreneurs have repeatedly been shown what happens when they appear too independent. Most recently, executives at the Chinese AI company Manus were prohibited from leaving China pending a government investigation into Meta’s acquisition of the company. (And the government eventually blocked the deal on national security grounds.) Jack Ma’s five-year disappearance from public view after criticizing financial regulators remains the most famous example. It’s not uncommon for Chinese authorities to “disappear” business executives, a practice that has increased under Xi: Some are never heard from again; some reappear and quietly return to work; some are prosecuted and imprisoned; and some even die under mysterious circumstances while in custody.

The appearance of order in China’s AI discourse is therefore not simply the result of public enthusiasm and technological pragmatism. It is also the product of a political system in which the state defines the strategic purpose of AI, sets the boundaries of permissible debate, constrains civil society, and reminds technology companies that they operate at the communist party’s pleasure.

For U.S. policymakers, this means cooperation with China on AI guardrails should be approached with a clear-eyed understanding of Chinese state control. The two countries are not only competing over models, chips, and talent. They are operating within different AI governance ecosystems. Chinese AI companies do not operate as independent actors in the way their American counterparts do, and the absence of visible anti-AI resistance in China hardly reflects genuine public consensus. Any U.S.-China AI negotiation that treats Beijing’s stated priorities as representing “China’s” full range of concerns risks reinforcing the very system that suppresses those voices.