In Jiashan county, Zhejiang, a man identified as Shi is suing Beijing Chuntian Zhiyun Technology, the provider of the Doubao AI chatbot, over burial‑date advice he followed for his mother’s funeral. The chatbot first recommended April 19 as the most suitable date, then told him the same date was not auspicious after relatives had been informed. A relative later suffered serious injuries in a traffic accident, some family members blamed the burial date, and Shi filed a network‑tort lawsuit seeking an apology and unspecified damages. The Jiashan County People’s Court held a hearing on September 3.
The case poses a question the court has not yet settled: whether a chatbot’s guidance on a culturally sensitive life event can expose its maker to legal liability when the user agreement calls all AI content “for reference only.” The Supreme People’s Court issued national guidance on AI disputes days later, directing judges to weigh how clearly platforms warned users about limitations.
Shi used the Doubao chatbot for everything. He organised work contracts with it, asked it daily questions, and when his mother died suddenly on April 17, 2026, he asked it for a burial date. It gave him one date, then told him it was wrong after relatives had already been told. Then it helped him draft the lawsuit against its own maker.
The youngest of five siblings in a county where the traditional almanac still governs the most important days of a life, Shi had reason to get the date right. The family first consulted a feng shui master and chose April 20. Local custom treats even‑numbered days after a death as unlucky, and April 20—the fourth day after his mother’s death on April 17—fell into that category. The chatbot recommended April 19 instead, specifying the burial window—7am to 9am—and the paper offerings to burn. After plans were fixed, it added that April 19 was “not auspicious.” A relative’s subsequent road accident turned family strain into a court filing. The tool that produced the contradictory advice also generated the legal papers to challenge it.
The machine said one thing, then another
Reports by Zhejiang Jiaxing City News Media Center’s Xiaoxin Shuoshi programme laid out the sequence. Shi’s mother died on April 17. The family consulted a feng shui master, chose April 20, then grew uneasy because the fourth day after death—an even number—is considered inauspicious in parts of the region. Shi opened Doubao, which recommended the 19th and detailed the ritual: burn paper horses and paper money before the procession, hold the burial between 7am and 9am.
After relatives had been notified of the new date, the chatbot added that April 19 was not suitable. Shi could not change the plans. Shortly afterwards, a relative was seriously hurt in a traffic crash. Some family members drew a line between the disturbed feng shui and the accident. The blame for that fell squarely on the youngest sibling.
The Jiashan County People’s Court opened a hearing on September 3, classifying it as a network-tort liability dispute. Shi seeks an apology and damages he has not made public. Days later, on September 7, the Supreme People’s Court issued its first comprehensive opinions on AI disputes, telling judges to apply the Civil Code’s Article 1165 fault-liability standard. Courts must examine whether providers gave clear warnings and took reasonable steps to prevent harm.
A lawyer cited by the Xiaoxin Shuoshi programme cautioned that an incorrect answer alone does not establish fault under China’s civil framework. The plaintiff must prove the provider committed a culpable act and that this fault directly caused his damage—a chain that remains in dispute. Another practitioner, Deng Yile of Beijing Xingquan Law Firm, argued that where a platform warns prominently that AI content is for reference only and works to improve accuracy, courts are inclined to find no liability.
This lawsuit arrives just as Doubao has scaled to hundreds of millions of users. The burial‑date case gathers those anxieties into a single court record.
The framework that absorbs risk
China’s legal architecture for AI is designed to let platforms grow while keeping liability inside narrow bounds. The Generative AI Service Management Interim Measures, in force since August 2023, treat AI outputs as services governed by network and data rules, not as products subject to stricter product‑liability law. The Supreme People’s Court’s September opinions sharpen this further: providers are liable only when they breach a duty and that breach causes damage.
Chinese legal commentaries on the Doubao dispute and on the Hangzhou Internet Court’s earlier AI‑hallucination ruling agree on the threshold: a claimant must show fault and a clear causal chain. In the Hangzhou case, judges found no liability where the platform had been open about the content being for reference only and the user could not prove actual damage. The Doubao chatbot’s own terms state the same.
Shi’s counter‑argument—that warnings were not prominent enough and that the company did not review the data the AI used—tests the limits of that defence. His reliance on the same tool to write the legal filings makes the asymmetry hard to ignore. It gave him a burial date, called it wrong, then helped him argue this constituted fault. All from the same server.
The Jiashan court’s ruling, when it arrives, will test whether the “service” label holds when cultural expectations and family harm attach to the output. If the court finds even limited responsibility, Chinese platforms face a redesign of their warnings. If it rejects liability, the “reference‑only” model holds—and millions of users who treat these tools as everyday counsel will remain exposed in ways the law still does not address.
Beyond the headline
The Power Behind It
Behind the burial‑date dispute is China’s effort to keep control of AI liability within established civil‑law boundaries while allowing platforms backed by ByteDance to keep scaling. By framing AI advice as a “service” governed by fault‑based tort rules and consumer‑law constraints, regulators preserve room for innovation yet implicitly shift ultimate responsibility for everyday decisions back onto users. The case exposes how platform design choices around disclaimers and access to self‑help legal tools quietly shape who bears the consequences when cultural norms collide with probabilistic outputs.
The Science Gap
Generative chatbots such as Doubao operate as probabilistic language models, not systems that understand feng shui principles or causal links between dates and misfortune. Chinese courts increasingly acknowledge phenomena like “AI hallucination” as inherent limits of current technology, but users often interpret fluent answers as authoritative guidance on traditions and risk. The gap between algorithmic pattern‑matching and genuine cultural expertise is where disputes like this emerge: people seek certainty from systems built to estimate plausible responses rather than to validate the underlying beliefs they reference.
What Isn’t Being Said
Coverage of the Doubao lawsuit largely focuses on superstition versus technology, yet says little about socio‑economic dependency on free consumer AI. The plaintiff’s reliance on Doubao for contracts and even litigation reflects a growing population that cannot readily access paid professional advisors but can summon AI help instantly. That structural reliance makes disclaimers about “reference only” use less realistic in practice, because for many users AI is the only convenient counsel they have. Recognising this gap would force a harder conversation about whether low‑cost mass AI tools now perform de facto advisory roles that regulation still formally denies.
A county court ruling with reach
With the Jiashan court’s decision expected in the coming months and China’s AI liability framework still taking shape, four groups watch for signals.
- Western consumer of AI chatbots for personal advice
Your own reliance on chatbots for legally or culturally significant decisions carries a risk that disclaimers may not be tested in your jurisdiction the way they are being tested now in China. Review the user agreements on services such as ChatGPT, Claude, or Gemini, noting how they frame advice as non‑professional and what responsibilities they assign to the user. For decisions with real consequences, consult a qualified local professional—a lawyer, a financial adviser, or a cultural authority—rather than treating an AI output as stand‑alone counsel.
- Legal professional advising on AI product liability
The Jiashan ruling will offer one of the earliest applications of China’s new Supreme People’s Court opinions to a consumer AI dispute involving cultural harm, and it may influence how fault and causation are interpreted in network‑tort claims globally. Monitor the Jiashan County People’s Court docket or coverage from the Xiaoxin Shuoshi programme for the judgment text, and watch whether the reasoning addresses the conspicuousness of warnings or the review of underlying training data.
- Product manager for consumer‑facing AI services
Even if the court rules for the defendant, the reputational cost of a case like this is a signal to audit your product’s disclaimers, user onboarding, and content moderation for advice touching on cultural practices. Ensure that important warnings appear prominently—not just in a scroll‑through agreement—and consider whether your moderation tools can flag queries that invite the system to opine on high‑stakes personal decisions.
- Ethicist or researcher in AI and cultural impact
This case crystallises a tension that existing AI‑ethics frameworks often overlook: the harm that arises when probabilistic language generation collides with deeply held cultural beliefs. The plaintiff’s dual use of the same chatbot—for ritual guidance and for litigation—makes the incident a rich case study for how AI design can unintentionally disrupt social obligations and where current liability models fall short.
Explainer
- Doubao AI
- A generative‑AI chatbot developed by Beijing Chuntian Zhiyun Technology, a company wholly owned by Beijing Douyin Information Service and ultimately controlled by ByteDance founder Zhang Yiming. Launched in 2023, it has scaled to hundreds of millions of users in China and is used for tasks ranging from drafting contracts to answering cultural questions. Douyin Group Vice President Li Liang publicly advised in 2026 that its outputs be treated as references, especially for personal‑safety matters.
- Feng shui
- An ancient Chinese practice concerned with harmonising individuals with their surrounding environment, often invoked when choosing burial sites and dates to ensure good fortune for descendants. It draws on principles of energy flow, directional orientation, and astrological timing, and remains widely consulted in rural and urban communities for major life events. In this case, the family believed a misaligned burial date disrupted feng shui and contributed to a relative’s accident.
- Jiashan County People’s Court
- The basic‑level court in Jiashan county, Zhejiang province, which held a hearing on Shi’s network‑tort lawsuit on September 3, 2026. County‑level courts in China handle the majority of civil disputes, and their rulings can be appealed to the intermediate court. A judgment here that assigns even limited AI liability for cultural advice would ripple upward through higher courts and regulatory agencies.
- Fault‑liability principle
- The legal standard in China’s Civil Code, Article 1165, which holds a party liable for harm only when they breach a legal or contractual duty and that breach directly causes damage. The Supreme People’s Court’s 2026 opinions direct judges to apply this standard to AI disputes, examining factors such as the clarity of risk warnings and whether providers took reasonable steps to prevent harm. It means incorrect AI output alone is not enough to establish liability.
RELATED STORIES
