China didn’t pass a new law. It applied an old one.
On May 2, a Hangzhou court ruled that a tech company’s dismissal of quality assurance supervisor Zhou, whose role had been automated by LLM technology, was unlawful under China’s Labour Contract Law. The company had offered Zhou a pay reduction reportedly described by some outlets as approximately 40% after automating his role. He refused. They dismissed him. The court found that dismissal unlawful, and in doing so, the court applied existing statutory language to a question no legislature had yet answered: does deploying AI to replace a worker give an employer the right to terminate that worker?
China’s answer, delivered through litigation rather than legislation, is no.
The Zhou Case: What the Ruling Actually Said
The facts are specific. Zhou worked at an unnamed Hangzhou tech company from November 2022, supervising quality assurance processes. The company adopted LLM automation that rendered his role redundant, by the company’s own account, not in dispute. Rather than reassigning Zhou, the company offered him a reduced role at substantially lower pay. He declined. The company terminated his employment.
According to NDTV’s reporting, the court found that AI automation does not constitute a major change warranting dismissal, characterizing it as a business decision rather than a legal necessity. The more specific legal doctrine, whether the court used the precise “impossibility of contract performance” framing, is characterized in some reporting but not independently confirmed in the available source material. What is confirmed across multiple outlets: the court found that voluntary AI adoption does not meet the threshold for contract termination under the Labour Contract Law, and the employer needed to demonstrate genuine operational difficulty, not merely a strategic preference for automated tools, to justify the dismissal.
A similar ruling is also cited from Beijing courts. The details of that ruling are not independently confirmed in the available source material, and should be treated as “at least one confirmed ruling from Hangzhou, with reporting also citing a Beijing case.”
China’s Legal Architecture: Why Case Law Moves Fast Here
China’s Labour Contract Law (2008, amended 2012) contains specific provisions governing when employers can lawfully terminate contracts. The law isn’t silent on workforce restructuring, it requires employers to demonstrate specific conditions (material changes in objective circumstances, genuine operational necessity) before terminating workers unilaterally. What the Hangzhou court did was apply that existing framework to an AI-driven dismissal.
In China’s civil law system, court rulings don’t operate as binding common-law precedent the way they do in the UK or US. But consistent judicial outcomes, particularly across Hangzhou and Beijing courts, create de facto standards that employment attorneys advise clients to treat as authoritative. And if China’s Ministry of Human Resources and Social Security issues guidance formalizing this standard, which is a real possibility following sustained judicial consistency, it converts from persuasive case law into administrative policy. That’s the escalation path to watch.
The Comparative Frame: California and the EU
China’s ruling didn’t emerge in a vacuum. Two other major jurisdictions are working toward their own answers to the same question, through very different mechanisms.
California’s “No Robo Bosses” Act takes a prospective legislative approach. The bill requires employers to disclose when automated systems are making employment decisions and gives workers the right to human review of AI-driven adverse actions. It does not categorically prohibit AI-driven layoffs, it requires transparency and appeal rights. The trigger is the automated decision itself, not the employer’s reason for adopting the tool. An employer in California could, in theory, use AI to eliminate a role and proceed with the dismissal as long as the worker received proper disclosure and the opportunity for human review.
The EU’s approach is structural. The EU AI Act classifies AI systems used in employment and workforce management, hiring, promotion, task allocation, and termination monitoring, as high-risk systems under Annex III. High-risk classification requires conformity assessments, technical documentation, human oversight measures, and transparency obligations. The Act doesn’t prohibit employers from using AI in workforce decisions, but it imposes significant process requirements on how those systems are deployed and governed. The August 2, 2026 deadline makes these requirements binding for Annex III deployers in 89 days. The Act doesn’t directly regulate the dismissal itself, it regulates the system used to make or influence the decision.
| Jurisdiction | Trigger | Standard | Employer Obligation | Enforcement |
|---|---|---|---|---|
| China (Labour Contract Law + May 2 ruling) | AI-driven role elimination + dismissal | Employer must prove “genuine operational difficulty”, strategic automation choice is insufficient | Reassignment or demonstrated necessity before termination | Civil litigation; administrative guidance possible |
| California (“No Robo Bosses” Act) | Automated system influences adverse employment action | Transparency and human review rights required | Disclose automated decision-making; provide human appeal path | State labor enforcement; private right of action |
| EU (AI Act Annex III) | Deployment of AI system in employment/workforce management decisions | Conformity assessment and human oversight required for the system itself | Technical documentation, HRMS audit, transparency to workers | EU AI Office; national market surveillance authorities |
The asymmetry is significant. A multinational deploying AI to manage workforce decisions must simultaneously satisfy three different frameworks that are triggered differently, impose different standards, and are enforced by different authorities.
Stakeholder Map: Who Wins, Who Loses, Who Must Act
*Tech companies with China operations* face the most immediate impact. If they’ve communicated to workers that AI automation eliminated a role, that communication is now potential evidence in a wrongful dismissal claim. The documentation trail matters enormously.
*Multinational HR teams* managing global frameworks face an architecture problem. A single workforce restructuring policy built for California or EU compliance doesn’t transfer to China. The legal standard is different, the trigger is different, and the remedies are different. Global HR frameworks need jurisdiction-specific provisions, not a universal policy with carve-outs.
*Workers in AI-adjacent roles in China* have stronger protection than their counterparts in most other jurisdictions, at least on paper. The Hangzhou ruling gives them a viable litigation path. Whether that path is accessible to most workers in practice is a separate question.
*Compliance and legal teams* at companies considering AI-driven workforce restructuring in any of these three jurisdictions now face a genuine multi-framework analysis requirement. The same restructuring plan needs three separate legal assessments.
What Compliance Teams Should Do Before This Spreads
Five concrete actions, grounded in what the ruling requires:
First, audit existing AI-driven workforce decisions in China-operating entities. If any dismissals in the past 12 months cited AI automation as the reason, formally or informally, assess exposure under the Hangzhou standard.
Second, review China-based HR communications protocols. Operational rationale language in termination documentation should be reviewed by employment counsel familiar with China’s Labour Contract Law framework. What was prudent to document six months ago may now be legally risky.
Third, do not assume the Hangzhou standard is limited to quality assurance or tech sector roles. The Labour Contract Law provisions being applied are general, not sector-specific. Any role eliminated in connection with AI automation is potentially in scope.
Fourth, watch for APAC jurisdiction spread. The Hangzhou ruling follows similar logic that employment law advocates in South Korea, Japan, and Australia have been advancing. If this doctrine reaches other APAC jurisdictions with strong labour protections, the compliance geography expands significantly.
Fifth, if your organization is implementing EU AI Act compliance for Annex III employment systems before August 2, the human oversight documentation you’re creating for EU compliance may also be your best evidence of procedural good faith if a China-based dismissal is challenged.
The question worth asking before your next AI-driven restructuring decision in China: can you demonstrate genuine operational difficulty, not just strategic preference for an automated tool, in writing, without relying on the fact of automation itself as the justification?