China’s Supreme People’s Court has issued guidance for cases involving deepfakes, voice cloning, algorithmic price discrimination, and AI-generated false information, placing judicial remedies alongside the country’s growing set of AI rules. Xinhua reported that the guidance was released on September 7 and addresses how courts should handle disputes created by increasingly capable AI systems. The document is not a new national statute, but it tells judges how to apply existing legal principles to specific AI-related conflicts.
The guidance matters because AI systems now create evidence and harms that are difficult to categorize through older legal labels. A cloned voice can be made from a short recording. A recognizable digital replica can circulate before a victim has time to respond. A pricing algorithm can offer different terms to different customers without explaining the basis. The court’s approach is to identify the parties who can bear responsibility in those situations: providers, businesses, and users.
This is a judicial development rather than an isolated enforcement campaign. It builds on the broader regulatory path EastFrontier covered in China’s digital-human draft rules and the later AI-misuse enforcement campaign. The new guidance supplies a more detailed view of what may happen when AI-related harm reaches court.
Digital Replicas Require Consent and Accountability
Xinhua’s account of the guidance says that creating or distributing a recognizable AI-generated digital replica without consent can infringe an individual’s rights. The examples include cloned faces and voices. The standard is important because it focuses on whether a person is recognizable, rather than only on whether an image was technically manipulated.
The court’s position does not mean every synthetic image or voice will automatically be unlawful. Context, consent, and the way material is used remain relevant. A person may authorize a digital likeness for a film, advertisement, service, or creative project. The legal risk arises when recognizable identity traits are used without permission in a way that infringes rights.
Xinhua connected the new guidance to an April 2024 decision by the Beijing Internet Court involving an identifiable AI-generated voice. The earlier case established that unauthorized reproduction of such a voice could infringe a person’s rights. The Supreme People’s Court guidance now brings that kind of dispute into a wider framework that also covers false information and consumer harm.
For people harmed by synthetic content, the guidance also identifies a route to urgent relief. Xinhua said a court may issue an injunction where AI-generated material threatens serious and difficult-to-repair damage. The report gives the example of a digitally altered face used to spread defamatory sexual claims. An injunction is not guaranteed in every case, but the inclusion of this mechanism signals that courts may intervene before the full merits of a dispute are resolved when delay would be especially damaging.
Algorithms Can Create Consumer Liability
The new guidance also addresses algorithmic price discrimination. Businesses that use algorithms to offer different prices or transaction terms for identical goods or services without a reasonable justification can be held liable when consumers are harmed, according to Xinhua.
This is a significant move because discriminatory pricing can be difficult for a single customer to detect. A consumer may see only the price displayed on one account, device, or time of day. The court guidance recognizes that an algorithmic system can become part of the evidence and liability analysis, even when a human employee did not personally set each price.
The court’s framing does not ban all differentiated prices. Businesses may offer different terms for legitimate reasons such as location, timing, membership benefits, or a clearly defined promotion. The question is whether comparable customers receive different treatment without a reasonable basis and suffer harm as a result. That distinction will matter when judges apply the guidance to actual facts.
The development follows a longer pattern of Chinese legal institutions adapting to AI-related complaints. EastFrontier reported in June that Chinese AI court cases were increasing as policymakers identified legal gaps. The Supreme People’s Court guidance does not resolve every gap, but it provides a more direct judicial vocabulary for identity, consumer, and content disputes.
Providers and Users Both Face Duties Over AI Output
The court guidance distributes responsibility across more than one actor. Providers may be liable when they are notified that their system has produced content infringing another person’s rights and fail to act promptly. Users may also incur liability when they intentionally design prompts or workflows to make an AI system produce rights-violating material that causes harm.
That division is important because generative systems can produce harmful material through many routes. A provider may build the system, a business may embed it into a service, and an individual may use it to create a specific falsehood. The court is not treating responsibility as belonging automatically to only one party. Instead, it asks what each party knew, what it controlled, and how it responded.
Xinhua also says the guidance covers AI hallucinations, meaning false or inaccurate output. The legal significance is not that every factual error by a model will lead to liability. AI systems can make mistakes in ordinary interactions. The guidance focuses on situations in which false output infringes rights and a provider fails to take timely action after notice.
This approach will likely increase the value of records showing how a model was used, what notice a provider received, and what corrective action followed. It may also encourage companies deploying AI systems to build clearer reporting, moderation, and appeal processes. Those operational responses are not described as mandatory technical requirements in Xinhua’s report, but they follow logically from a framework that makes notice and response important.
China’s AI legal system now includes labeling rules for generated and synthesized content, policy measures for digital humans, enforcement actions against misuse, and court guidance for disputes. The Supreme People’s Court has added a judicial layer focused on remedies and responsibility. The next test will come in individual cases, where courts will have to translate general principles into decisions about consent, deception, pricing, and provider conduct.
For users, the guidance offers more explicit grounds to challenge unauthorized replicas and harmful synthetic content. For companies, it signals that AI products cannot be evaluated only by capability and adoption. The way they handle notices, consumer treatment, and high-risk output may become central to legal exposure. That is a material shift for a market in which AI systems are moving from demonstration to daily use.