Beijing forbids firing workers over AI. London allows it.

15,000 yuan a month. That's what Zhou's employer wanted to impose after AI absorbed part of his job. A third less than his previous salary. Zhou refused.

He took it to local arbitration. He won at first instance. On April 28, 2026, the Hangzhou Intermediate People's Court upheld the ruling on appeal.

And it isn't the first decision of its kind. It's the second. Between the two, you have a body of case law forming.

The Zhou case in four acts

Zhou worked at a tech company in Hangzhou, Zhejiang. His role: quality assurance supervisor, checking that the outputs of the LLMs used by the company were correct. Base salary, 25,000 yuan a month, roughly $3,640.

As AI rolled out across the company, his employer offered him a demotion to 15,000 yuan. Zhou refused. The employer terminated the contract, citing the disruptive impact of AI on the position.

The local tribunal ruled the dismissal unlawful. The company appealed. On April 28, 2026, the Hangzhou Intermediate Court confirmed. AI alone, as a reason, isn't enough.

The legal detail that matters

Chinese labor law allows an employer to terminate a contract in the event of a "major change in objective circumstances" that makes the working relationship impossible to continue. That's Article 40 of the Labor Contract Law. The concept covers external shocks: disasters, bankruptcy, major regulatory shifts.

On this point, the court is precise. Picking up the reasoning from a Beijing arbitration panel already cited in December 2025, it states that "the adoption of AI may require adjustments to the structure of jobs, but these adjustments fall within the risks an employer should reasonably anticipate as part of normal business operations." Translation: deploying AI is a strategic choice by the company, not an event being suffered. You don't push that choice onto the worker.

From precedent to doctrine

The precedent Hangzhou refers to is called Liu. Hired in 2009 at a mapping company to enter data manually, Liu found himself without a position in early 2024 when his employer switched to AI-based data collection and eliminated his division. Dismissed in late 2024, he won in arbitration. The Beijing Municipal Human Resources Bureau published his case in December 2025 among the ten most significant labor arbitration decisions of the year.

Two rulings, two regions, five months apart, same logic. What's emerging now is a guideline, not an isolated case.

And the timing isn't neutral. According to figures compiled by The Next Web, more than 78,000 tech jobs have been cut worldwide between January and April 2026, with roughly half explicitly attributed to AI replacement. Beijing is signaling to Chinese companies: we're watching.

The other side of the mirror

Here's where the contrast becomes uncomfortable. Section 139 of the UK Employment Rights Act 1996 defines redundancy as a situation where "the requirements of the business for employees to carry out work of a particular kind have ceased or diminished, or are expected to cease or diminish." The text is explicit. If AI takes over the work, the legal mechanism is open.

The application is straightforward. Introducing AI to process invoices, draft notes or moderate content fits the definition directly. UK employment specialists have made it explicit. Pearce Legal addresses it as a routine question: "Can I replace staff with AI and make them redundant?". The answer, in legal terms, is yes.

The only safeguard is a fair process. The employer has to consult, apply objective selection criteria, respect statutory notice. If the process is followed, AI as the underlying reason holds up. There's no AI-specific statute, no doctrine that blocks the use of AI as a redundancy trigger. Whether or not a major UK case has crystallised the question yet, the legal route is wide open.

In China, that route just got narrower. In the UK, it's a road employers can take, provided they handle the process correctly. The mechanism stays open. Hangzhou just locked it.

The invisible regulation

This is where it gets geopolitically awkward. When we talk about AI regulation in 2026, we look at the EU AI Act, US executive orders, the policy declarations from Bletchley or Seoul. We don't look at Beijing. The dominant narrative places China on the side of technological laissez-faire and state capitalism indifferent to individual rights.

Except that on this specific point, Chinese law has just drawn a line that neither the AI Act nor UK employment law draws. The AI Act handles model transparency and systemic risk, not employment protection against automation. UK employment law recognises AI-driven redundancy as a legitimate route. The Hangzhou court, by contrast, refuses to treat AI alone as a major objective event.

That's a direction, not a technical detail. If China maintains this doctrine, its tech companies will have to factor the cost of the human transition into their AI deployments: redeployment, training, negotiation. Either they slow down, or they find formal workarounds (negotiated demotions, transition programs). Either way, the trade-off between productivity and social protection is being made over there.

The AI regulation we're hunting for in European parliamentary chambers just played out in a Chinese labor court. And nobody in Europe was watching.