I recently learned of a high-profile case involving Large Libel Models in China, adding to the set of nine U.S. cases I’ve encountered and alongside other examples from Norway and Germany. The following is an excerpt provided by Mengchun Zhang (China IP Case Updates) of JunZeJun Law Offices, dated May 24, 2026, summarizing a decision issued on March 12, 2026.
Nanjing Intermediate Court holds AI-driven claim that a lawyer received a three-year prison sentence to be defamatory …
The court drew a line between routine search suggestions and content produced by AI. While general search features like “related searches” were not found to infringe, the court determined that the relevant “AI Smart Answer” material originated from Baidu’s AI processing and the synthesis of text and images.
The ruling is significant because the court still found Baidu liable despite its arguments about AI hallucination and the limitations of technology.
At the same time, the judgment did not provide further elaboration on the precise standards for attributing platform fault or the scope of platform duties in AI hallucination scenarios….
The Nanjing Jiangbei New Area People’s Court ordered Baidu to issue a formal written apology to Li Xiaoliang.
The court did not sustain Li’s claims for economic damages or for emotional distress, because he failed to present sufficient evidence of such losses.
Baidu appealed the decision.
In March 2026, the Nanjing Intermediate People’s Court affirmed the first-instance judgment.
After the judgment became effective, Baidu did not voluntarily comply, and Li Xiaoliang subsequently filed for enforcement with the court.
I contacted Mengchun Zhang to inquire about any further developments, and he kindly replied that “the enforcement division of the court informed Mr. Li that the enforcement case would soon be formally accepted for enforcement,” signaling that additional developments may be on the horizon.
