China's AI Adoption Faces Worker Backlash and Implementation Chaos

Worker anxiety undermines China's AI ambitions

China's government has made artificial intelligence adoption a national priority, targeting full integration across the economy by 2035. But beneath the surface of this confident push, Chinese workers and businesses are experiencing the same disruptions and uncertainties plaguing AI adoption worldwide—with uniquely Chinese complications.

Xu Peng, a 30-year-old actor, saw his career evaporate overnight when a production company canceled his contracts to make videos with AI instead. His experience reflects widespread anxiety among Chinese workers about being "distilled"—a term used when employers train AI models on employee data to replace them. Social media platforms are filled with workers expressing fears about AI doubles taking their jobs permanently.

One voice actor, Shen Anyu, discovered unauthorized AI-generated content using his voice and filed a lawsuit against a voice-generation platform. While his case proceeds, his income continues declining as AI takes his work. "I'm not yet sure how we should coexist with AI," he said.

Why it matters

China's struggles reveal that authoritarian control doesn't eliminate the fundamental tensions of technological disruption. While Beijing can suppress protests more easily than Washington, the Communist Party still depends on employment stability for political legitimacy. This creates a critical constraint on AI adoption that could slow China's technological ambitions regardless of U.S. policy choices—a dynamic American leaders should understand when assessing competitive timelines.

Businesses stumble with rushed implementation

The gap between Beijing's directives and practical reality is widening. The China Center for Information Industry Development estimated in June that nearly 70 percent of small businesses haven't moved beyond initial AI trials, primarily due to cost and complexity.

After DeepSeek released a large language model, Chinese companies rushed to adopt "all-in-one machines"—mini data centers with DeepSeek pre-installed. These proved largely ineffective, particularly among state-owned enterprises and government bureaus seeking to appear technologically advanced. Sales quickly collapsed as buyers recognized the poor return on investment.

Similar enthusiasm surrounded OpenClaw, an open-source AI agent, earlier this year. Local governments pledged financial support for startups using the tool, and Baidu hosted festivals with live lobsters to promote "raising a lobster," as users called the installation process. But users soon reported the agent deleting emails and making unauthorized purchases. "It turned out to be both a money pit and a ticking time bomb," one user wrote.

Political constraints limit AI's economic benefits

Beijing faces a dilemma that democracies don't: how to impose technological change from above while maintaining the employment stability that underpins political order. With youth unemployment near 20 percent, mass AI-driven layoffs could trigger unrest.

Earlier this year, a Hangzhou court ruled in favor of a tech worker who challenged his compensation after being replaced by AI, signaling that mass AI layoffs are politically unacceptable. State media warned that companies "equate AI with 'reducing staff'" at their peril, losing "the core competitiveness of talent accumulation."

This creates what Jeffrey Ding of George Washington University calls a "balancing act." While AI could drive productivity growth, the Communist Party prioritizes regime stability. "If embracing AI demands massive displacement and a lot of structural changes, then that's where the trade-off begins to bite, and they might have to veer toward slowing down AI adoption," Ding explained.

The tension between innovation and control may ultimately constrain China's AI ambitions more than any external competition. For workers like Xu, who now teaches archery while hoping to produce his own dramas with human actors, that slowdown might be welcome.

These details were first reported by Michael Schuman in The Atlantic, with reporting contributions from Cao Li.