Chinese courts rule on AI platform liability for defamation and copyright infringement
Chinese courts ruled in two separate cases on AI platform liability. In one case, Baidu's AI system generated false information describing a lawyer as convicted of a crime; a court found this constituted defamation and ordered an apology though rejected damages. In another case, a court found an AI search platform not liable for pirated content links appearing in results, determining the platform fulfilled its legal obligations through complaint mechanisms and timely removal.
What happened
Baidu's AI Smart Answer feature generated false information describing Nanjing lawyer Li Xiaoliang as having received a three-year prison sentence, displayed alongside his photograph in search results. Li Xiaoliang discovered the false content in 2024 on Baidu's mobile app and website.
A media company discovered that searching for television drama titles on an AI search platform returned pirated streaming links positioned at the top of results, with text such as 'no extraction code needed.'
Reported impact
Li Xiaoliang's social reputation was damaged by the false defamatory content generated by Baidu's AI system, affecting his professional and personal life.
The pirated links in the AI search results infringed the media company's information network distribution rights for the copyrighted television dramas.
Index analysis
Nanjing Jiangbei Xinqu Court found that while Baidu's regular search suggestions and related searches derived from user search records without direct processing, the 'AI Smart Answer' feature involved AI technology processing and generation of content explicitly indicating negative information about Li Xiaoliang. The court determined Baidu was negligent despite claiming AI hallucinations are unpredictable and accidental.
Shanghai Xuhui District Court found that the AI search platform, based on large language models and retrieval-augmented generation technology, cannot avoid indexing and displaying content from public internet sources. The court determined the platform did not manually edit or recommend results and had no subjective fault, and that it had fulfilled its obligations by maintaining complaint channels and promptly removing infringing content upon notification.
Governance controls
Baidu appealed its first-instance judgment on grounds that AI systems cannot predict user queries or fully avoid hallucinations, but the Nanjing Intermediate Court rejected this argument, noting that other AI platforms such as Douyin and DeepSeek did not produce similar defamatory results.
The AI search platform demonstrated due diligence through model and algorithm registration, establishing complaint channels, and removing infringing links on the same day a lawyer's notice was received.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.