There is a story about China and AI that the West tells itself. It goes like this: Chinese citizens broadly accept AI because the state tells them to. Surveys show high approval. There is no public resistance. Therefore, China will move faster and with fewer constraints than democracies.
A recent article on Lawfare, “The Missing Resistance in China’s AI Debate” by Yaqiu Wang, makes this case powerfully. It documents censorship, suppression of dissent, and the systematic elimination of civil society opposition to AI. It is a compelling read. It is also only half the story.
Let me give you both sides.
The Case for the Prosecution
The Lawfare article builds a strong evidentiary record. A KPMG survey across 47 countries found that 69 per cent of Chinese respondents said AI’s benefits outweigh the risks, compared to just 35 per cent in the United States. That gap is not accidental, Wang argues. It is the result of state control.
Chinese law professor Lao Dongyan criticised facial recognition in Beijing’s subway and big data surveillance. Her social media posts were deleted, her accounts suspended, and state media attacked her. Similar criticism is now rare. In Shanghai at Halloween 2023, a man dressed as a surveillance camera was questioned by police. If protest against the surveillance state can only be expressed through costumes, and even that is suppressed, how can anti-AI sentiment emerge openly?
Wang also notes that Chinese AI company leaders are not independent actors. Manus executives were banned from leaving China pending a government investigation. Jack Ma disappeared for five years after criticising financial regulators. Tech analyst Afra Wang puts it bluntly: “The companies build. The state then decides what it has been built for.”
These are not isolated incidents. They are features of a system designed to eliminate the kind of civil society pushback that exists in Western democracies.
The Case for the Defence
The counterargument is not that China is a free and open society. It is that the narrative of manufactured consent misses something important: China’s AI strategy has delivered tangible results that benefit real people.
China’s “AI Plus” plan, launched in August 2025, is a ten-year roadmap to integrate AI across the entire economy. It is backed by sustained political support that does not get derailed by election cycles. The Made in China 2025 plan that preceded it was largely successful in achieving global leadership in electric vehicles, advanced rail and renewable energy. There is no reason to assume AI Plus will fail where its predecessor succeeded.
Chinese labs have produced genuine technical innovations. DeepSeek’s mixture-of-experts architecture, Alibaba’s Qwen models, and Moonshot AI’s million-token context windows are not simply copies of Western work. Chinese researchers represent up to half of NeurIPS attendees. The open-source models coming out of China, from DeepSeek and Alibaba, have surpassed their American counterparts in total downloads on Hugging Face. Airbnb’s customer service agent relies heavily on Qwen. Silicon Valley startups choose Chinese open models because they work well and cost less.
The Brookings Institution, in April 2026 testimony before the US House Select Committee, concluded that export controls can slow China down but are unlikely to halt its AI progress. Chinese domestic AI chips captured approximately 41 per cent of the local market in 2025, up from less than 10 per cent before 2023.
These are not the achievements of a system running on propaganda alone. They are the results of sustained investment, genuine talent, and a national strategy that treats AI as a public good, not just a commercial opportunity.
The Missing Middle
The truth, as always, sits somewhere in the middle. Both sides have a point.
The Lawfare article is correct that public opinion in China is shaped by state control. The absence of visible resistance does not equal genuine consent. When the United States negotiates AI guardrails with China, it must understand that the Chinese government’s stated positions are not the product of democratic debate. They come from a system where dissent is costly and sometimes dangerous.
But the counterargument is also correct. The Chinese approach to AI is not simply imposed from above. It is also adopted from below, because the tools work. AI-powered medical diagnostics reach rural communities. AI-optimised logistics keep supply chains moving. AI-driven manufacturing improves quality and reduces costs. People use these tools not because the state forces them, but because they deliver value.
The danger for Western policymakers is assuming that a lack of visible resistance means the Chinese public opposes AI. It is equally dangerous to assume they would resist if they could. The reality may be more uncomfortable: many Chinese citizens genuinely see AI as beneficial, and the state’s priorities align with their economic interests, even if the political system behind it is not one we would choose for ourselves.
“The question is not whether China’s AI consensus is real or manufactured. It is whether a system that suppresses dissent can still produce outcomes that people value. The evidence suggests it can, and that is the more challenging truth for the West to confront.”
What This Means
For anyone watching the US-China AI competition, both narratives matter. The suppression of dissent in China is real and documented. So is the technical progress. Dismissing either one leads to bad policy.
The winner of the AI race will not be determined by who has the most public support for AI. It will be determined by who can most effectively translate AI into broad-based economic and societal gains. On that measure, both the United States and China have genuine strengths, and neither should underestimate the other.
Related Reading
- Cloudflare Just Taught an AI to Chain Security Bugs Into Real Exploits
- Scammers Are Now Hosting Malware on chatgpt.com. Yes, the Real One.
- The AI Sandbox Myth: Why Your Security Tests Are Hacking Real Companies
The views expressed on this site are my own and do not represent those of any current or former employer. Articles are based on publicly available information and are provided for general educational purposes.
