China's Supreme People's Court has released the country's first judicial document on AI-related adjudication rules, a 24-article guideline that makes unauthorized voice cloning and AI-generated digital replicas an infringement of personality rights. The rules also set liability for AI providers that ignore infringement notices and for users who deliberately prompt infringing output.
On September 7, 2026, China's Supreme People's Court issued the 'Opinion on Lawfully Adjudicating Cases Involving Artificial Intelligence Disputes,' a 24-article guideline and the first judicial document on AI adjudication rules released by the country's top court. The guideline addresses the most common deepfake-related abuses — AI face-swapping, voice cloning, AI hallucinations, and AI-enabled doxing — and establishes that creating, using, or distributing recognizable digital replicas of a person without consent infringes their personality rights, while using someone's voice without consent to imitate their speech patterns violates voice rights. Using AI-generated faces or voices to spread false claims or damage someone's reputation also creates legal liability.
The document's liability framework reaches both sides of the AI stack. Providers can be held liable if they are notified that their systems generated infringing content and fail to take timely action, and users who deliberately craft prompts to induce infringing output also bear responsibility. Courts may grant injunctions when AI-generated content threatens serious and irreversible harm — for example, digitally altered faces used to spread false, defamatory sexual claims. The guideline further covers algorithmic price discrimination against consumers and 'box-opening,' or using AI to track and analyze publicly available personal data such as phone numbers and social media profiles to disturb an individual's peace. Si Yanli, deputy head of the Supreme People's Court's Research Office, underscored the stakes at the news conference: 'AI voice cloning can 'steal' someone's voice for just a few yuan, creating synthetic results that sound convincingly real.'
The rules build on China's September 2025 measures governing labeling of AI-generated content and the Beijing Internet Court's April 2024 ruling in the country's first AI-generated voice infringement case. They land amid explosive growth in AI-generated short dramas, where unauthorized face-swapping and voice cloning of celebrities has become widespread — lawyers told the Global Times the guideline establishes 'identifiable equals infringement' as a core adjudication standard for the first time, transforming previously vague administrative requirements into directly enforceable judicial rules. China's AI industry surpassed 1.2 trillion yuan (about $177 billion) in 2025, up 40 percent year on year, with more than 6,600 AI companies as of June 2026, roughly 15 percent of the global total.
Industry observers expect the rules to hit speculative, low-quality AI productions hardest while leaving legitimate studios — which rarely use unauthorized cloning — largely untouched. Analyst Zhang Peng told the Global Times that the guideline raises the cost of infringement 'from near zero to actionable and compensable,' and called the shift 'short term pain for opportunistic players and long-term gains for the whole sector.' Beijing-based lawyer Wu Xiaolin framed the broader significance: the rules 'offer a judicial expectation for the broader entertainment AI industry and provide a governance example for other countries.' For voice AI developers everywhere, the message is consistent with the EU AI Act's transparency regime: a person's identifiable voice is a protectable asset, and consent — not availability — is the test.
