(Yicai) Sept. 8 -- The Supreme People's Court of China has released its first judicial ruling document for disputes related to artificial intelligence, providing a basis for the adjudication of cases arising from the application of AI technology that infringe on personal, intellectual property, and consumer rights.

The document focuses on issues of great public concern, including AI-generated deepfakes, the use of AI to obtain and disclose personal information, algorithmic price discrimination, and traffic accidents involving autonomous vehicles, and clarifies the corresponding ruling rules, Si Yanli, deputy director of the SPC Research Office, said at a press conference yesterday.

AI deepfake technology is among the areas with the strongest public response, Si pointed out. To address this, the document said that using AI to replicate a natural person's name or likeness to generate identifiable virtual digital images for use or public disclosure without consent constitutes an infringement of the name and portrait rights.

Meanwhile, using a natural person's voice as training data to generate recognizable synthetic voices without consent constitutes an infringement of voice rights, per the document.

The document also addressed the behavior of AI “resurrecting the deceased," stating that the unauthorized creation or use of virtual digital images of deceased individuals with AI technology that results in the infringement of their name, likeness, or reputation, will be subject to legal consequences. Courts will support the deceased's close relatives in seeking civil liability from the responsible parties under relevant provisions of the Civil Code.

To protect personal privacy, the document said that tracking, analyzing, or obtaining private information about specific individuals using AI on publicly available information, such as phone numbers, online accounts, and social media, or disclosing and publicly sharing the obtained information in a way that infringes upon the tranquility of private life, should be recognized as an infringement of privacy rights.

On consumer rights protection, the document specified that if AI is used to impersonate celebrities for online livestreaming sales and constitutes fraud, consumers can seek punitive damages. If operators use algorithms to implement unreasonable differential treatment that results in harm, they will be liable.

In addition, the document proposed referencing “safe harbor rules” to define the liability boundaries of generative AI service providers to balance the protection of rights and the development of the AI industry. For instance, service providers can be exempt from compensation liability if they promptly delete content upon receiving a notice of infringement.

It is challenging for AI service providers to foresee the content that users input in advance, and they are unable to review each generated result individually. However, once infringing content is produced and the rights holder notifies them, they must take necessary measures under certain conditions. Therefore, the rationale for referencing and applying the "safe harbor rules" is justified, Si said.

Editors: Dou Shicong, Futura Costaglione