Baidu Held Liable in Landmark Verdict Over AI-Generated False Criminal Record

A recent AI hallucination verdict has set a precedent for similar violations.

Li Xiaoliang, a lawyer in Nanjing, Jiangsu Province, searched Baidu's AI-enabled service for his personal information and found that he "has been sentenced to three years' imprisonment for the crime of explosion," along with a photo of him in his judicial robe.

Li was shocked by the findings. He sued the platform for defamation.

After two trials, Baidu was found guilty of defamation on March 12 despite its claim that "AI hallucinations are unforeseeable." The platform was ordered to write an apology.

This landmark ruling defined AI technology's limits and online platforms' liabilities as well as Li's legal rights.

This verdict differed from China's first lawsuit against large AI model infringement.

In a case handled by Hangzhou Internet Court, a user named Liang searched for university admission information on an AI app and received inaccurate campus locations and enrollment policies. After being misled, Liang sued the AI service, but the claim was dismissed.

Two seemingly opposing verdicts are explicable. The one-on-one human-machine interaction error involving university-related information caused no actual harm to the user.

The rapid integration of AI into almost every aspect of social life has spurred an explosion in AI-generated content, some of which is created without legal or ethical constraints.

Baidu's assertion that AI hallucinations are unpredictable and that the industry is still evolving is unfounded, as technical constraints and legal accountability must remain distinct.

Legally, Baidu's response involved algorithmically generated output that included images. This output was clearly harmful.

More persuasively, the judge cut to the core during the court hearing: no other AI products of the same kind have made such serious defamatory claims. This sufficiently proves the issue lies not in uncontrollable technology, but in the platform's failure to fulfill review and risk-prevention obligations commensurate with industry standards.

AI's tendency to spew such nonsense can only be moralized by holding technology and service operators fully accountable for AI-generated content, from content review to risk prevention and damage containment.

In other words, platforms cannot claim impunity from such hallucinations and deviations.

Other platforms, as gatekeepers, must be reminded of their responsibility for AI-generated content.