China has spent years telling companies to put artificial intelligence to work. Now Beijing is preparing for the possibility that some of those workers will no longer be needed.

At some Chinese technology companies, managers are tracking employees' use of AI and incorporating the numbers into performance reviews. Elsewhere, companies are cutting contractors, hiring fewer graduates, or shrinking production teams as AI takes over work once done by people.

So far, the cuts have generally been quiet. That may be the point.

Beijing wants AI to raise productivity across the economy, from factories and logistics to healthcare and finance. But China is also dealing with weak consumer confidence and a difficult job market, especially for young people. A wave of highly visible AI-driven layoffs would work against both goals.

The government is starting to prepare for that collision. In July, China's human-resources ministry and three other government agencies ordered the creation of an AI employment-impact assessment system. The government plans to study how the technology affects jobs, steer innovation resources toward areas with greater potential to create employment, and expand AI training for workers.

The language is unusually direct. The goal, according to the government plan, is to reduce AI's "impact on employment."

That concern is arriving at a bad time. China's official unemployment rate for people ages 16 to 24, excluding students, jumped to 17.9% in July, its highest in 11 months. A record 12.7 million university graduates are entering the job market this year.

Many are looking for exactly the sort of entry-level white-collar work that generative AI can increasingly do.

Citi estimates about 70 million Chinese jobs, or nearly 10% of the total, face a high risk of displacement from AI. Workers in their 20s are even more exposed. Customer service, advertising, entertainment, basic computer programming, and other office work are among the areas where companies are already experimenting with replacing tasks rather than adding employees.

Cai Fang, a member of the Chinese Academy of Social Sciences and a prominent labor economist, described the trade-off in unusually stark terms.

"We welcome AI's unprecedented creative potential. Its unprecedented destructive potential, however, must be controlled and managed to the greatest extent possible," he said.

That doesn't mean AI will ultimately destroy more jobs than it creates. Justin Yifu Lin, a former World Bank chief economist, has argued that new employment opportunities will eventually outnumber the positions displaced by AI. But, he has warned, "the transition process will involve pain."

Chinese companies are already showing signs of managing that transition differently from their U.S. counterparts. Rather than announcing large AI-related layoffs, some are allowing head count to fall through attrition, cutting contractors, or simply hiring fewer people. Chinese courts have also ruled against employers in several cases in which workers were dismissed and replaced by AI.

JD.com offers a glimpse of how the politics could play out. The e-commerce giant is pouring money into warehouse automation, robotics, and AI. Founder Richard Liu has nevertheless said the company will protect front-line workers displaced by machines and retrain them for other jobs.

For investors, that distinction matters.

One of AI's biggest promises is operating leverage. Companies can produce more without adding workers, or produce the same amount with fewer of them. Alibaba Group Holding, Tencent Holdings, Baidu, JD.com, and other large Chinese companies should have plenty of opportunities to capture those gains.

But Beijing may be less comfortable with the second half of the equation. Large-scale layoffs would add pressure to an already weak labor market and could further undermine household confidence and spending.

That could put an unofficial political limit on how aggressively Chinese companies turn AI productivity gains into head-count reductions.

Employment isn't the only area where Beijing is discovering that successful AI adoption creates new problems.

On Sept. 2, China's internet regulator said a campaign against AI abuses had removed more than 5.6 million pieces of illegal or harmful information, penalized or closed 49,000 accounts, and taken action against roughly 2,400 websites and apps.

Authorities are targeting AI-generated false information, impersonation, fraud, harmful content, and abuses involving minors. Police said earlier this year that they had investigated more than 170 cases involving AI-generated online rumors in the first six months of 2026.

Those problems are unlikely to make China retreat from AI. Beijing wants the technology embedded across industry, public services, and everyday life.

The challenge is increasingly what happens after it gets there. China's first generation of internet regulation often arrived after industries had already grown enormous and their social problems had become difficult to ignore. Gaming and private tutoring showed investors how abruptly Beijing could act once it decided those costs had become too high.

With AI, policymakers are trying something different: encouraging the technology while attempting to anticipate its consequences.

That approach could spare investors some of the regulatory shocks of the past. It could also constrain some of the gains companies hope AI will produce.

Beijing wants the productivity. It just may not be willing to accept unlimited disruption to get it.

Write to editors@barrons.com