.webp)
China’s Supreme People’s Court has issued new China AI dispute guidelines covering AI-generated face-swapping, voice cloning, model training, generative AI services, algorithmic discrimination, and autonomous-driving accidents. Published on September 7, the document is the court’s first national judicial guidance focused specifically on disputes involving artificial intelligence.
The guidance, titled Opinions on Adjudicating AI-related Disputes According to Law, contains five sections and 24 articles. The official Supreme People’s Court release says it applies existing provisions of China’s Civil Code, Cybersecurity Law, Data Security Law, Copyright Law, Consumer Rights Protection Law, Personal Information Protection Law, and other relevant laws. China does not yet have a dedicated comprehensive AI law, according to the release.
China AI dispute guidelines address deepfakes and voice cloning
The court also addresses AI-assisted doxxing and privacy violations. Using AI to track and analyse publicly available phone numbers, online accounts, or social-media information in order to obtain, disclose, or misuse private information may constitute an infringement of privacy rights under the personal information protection law. The rules may also intersect with cybersecurity law and data security law where large-scale data processing or misuse is involved.
For organizations managing synthetic-media risks, Nemko Digital’s reporting on deepfake threats and digital trust provides related industry context.
Liability extends to providers, products, and data practices
The guidance adopts a notice-based approach for some generative AI disputes. When AI-generated content infringes personality rights and a rights holder provides a valid notice, a provider that fails to take necessary measures, such as stopping the generation of the infringing content, may face platform liability. Users who deliberately induce a system to generate infringing material may also be liable. These rules clarify how responsibility may be allocated among service providers, users, and developers in AI-related disputes.
The document provides that processing lawfully public personal information for model training within a reasonable scope, where the individual has not expressly refused, will generally not be treated as an infringement. Consent remains required where the processing has a major impact on personal rights. Organizations can compare this requirement with Nemko Digital’s ISO/IEC 42005 AI impact assessment framework, which examines impacts across an AI system’s lifecycle.
The guidance further distinguishes physical AI products from services for product-liability purposes. For autonomous and driver-assistance vehicles, manufacturers or sellers may bear responsibility where a product defect contributes to an accident. They may also face liability for false or misleading claims about a vehicle’s automation level, intelligence, performance, or intended use. Courts may require data controllers to provide relevant event records to establish what happened under applicable civil procedure law.
The rules may also inform disputes involving algorithmic price discrimination and other conduct addressed by consumer rights protection law or anti-unfair competition law. This is particularly relevant as the artificial intelligence industry expands the use of big data, AI models, and artificial intelligence technology applications.
Evidence and unresolved copyright questions
The rules also make technical documentation relevant to litigation. When developers rely on a non-infringement defence in an AI copyright dispute, they may be required to provide information about training-data sources, training process records, model operation, and supporting technical evidence. The court’s official explanatory Q&A states that the guidance does not resolve whether AI-generated content is copyrightable or how the unauthorised use of copyrighted works to train large models should be classified under copyright law.
The unresolved questions are important for open-source software developers, businesses using AI-generated content, and organizations conducting artificial intelligence technology research. They also leave open broader issues concerning creative contribution, substantive features, and whether particular outputs qualify for protection.
For businesses, the immediate compliance focus is therefore traceability: documenting consent and data-use decisions, identifying training data sources, assigning responsibilities among developers, providers, and users, retaining model and incident records, and maintaining procedures for rights-holder notices. Nemko Digital’s global AI regulations overview and ISO/IEC 42001 AI management systems guide provide relevant governance context for organizations operating across jurisdictions.
The court said future rules may develop as experience accumulates on issues where consensus has not yet formed. Organizations should monitor subsequent cases, judicial interpretations, and cross-border developments, particularly where AI systems combine synthetic media, personal data, intellectual property, or safety-critical functions. Courts may also refine judicial adjudication rules for artificial intelligence disputes, including questions of evidence, judicial order, and the responsibilities of natural and legal persons.
Earlier official court reporting on the draft guidance shows that the rules emerged from growing disputes involving data rights and AI-generated content. As AI technology development continues, the highest court may provide further guidance on other targeted disputes, including trade secrets, misuse of AI systems, and the high-quality development of artificial intelligence technology applications.
