China · Zhejiang · Platform Liability

Hangzhou AI Hallucination Liability Case

A general-purpose chatbot gave incorrect admissions information and generated a compensation promise. The court rejected the claim after assessing attribution, care, harm and causation.

Court or tribunal
Hangzhou Internet Court
Decision year
2025
Procedural status
First-instance judgment effective in 2025; court-origin case reporting

01

Factual background & dispute

  • Liang received inaccurate information about college applications.
  • The conversation generated compensation language; Liang sought RMB 9,999.
  • The provider supplied evidence of conspicuous warnings and reliability measures.

02

Core issues & judicial focus

  1. Attribution of an automatically generated promise
  2. The applicable liability standard
  3. Proof of fault, loss and causation

03

Judicial finding & holding

  • In this general-purpose chat, the generated promise was not the provider’s declaration of intent.
  • Applying fault-based liability, the court found reasonable care and no proven loss or causation.
  • The judgment became effective without appeal; company-authorised customer-service statements may require a different analysis.
Dates & procedural record
  • Court-origin report published; the decision remains recorded at year precision.

04

Practical risk implications

01Separate general conversation from customer service authorised to communicate company commitments.

02Use prominent, timely warnings in high-impact contexts and retain reliability-testing evidence.

03Provide verification, correction and escalation processes, with complete conversation records.

PROJECT ENQUIRY

How does this precedent affect your product or litigation?

Consult with our practice team →