China — AI × Entertainment Regulation¶

⚠️ Not legal advice. → Comparison by country / Cross-cutting overview (Japanese) The original texts from Chinese authorities (CAC) have not been checked. The following is based on commentary from law firms and others (verification level B). Japanese edition: ../../../ethics-legal/by-country/china.md

Training¶

Interim Measures for the Management of Generative AI Services (promulgated July 13, 2023; effective August 15) Article 7: in training (pre-training and fine-tuning), providers must
use data and foundation models from lawful sources
not infringe others' intellectual property rights
obtain consent where personal information is involved
improve the truthfulness, accuracy, objectivity, and diversity of training data

Voice rights¶

Civil Code Article 1023: protection of a natural person's voice applies the portrait-right provisions mutatis mutandis. No one may distort, forge, or use another's voice without permission by means of information technology
Beijing Internet Court (April 23, 2024): China's first ruling on personality-right infringement by an AI voice. It held that the defendants' unauthorized use, in a text-to-speech app, of an AI voice built from the plaintiff's voice was infringing, and awarded a total of ¥250,000 (RMB). Protection extends to AI-generated voices as long as they are identifiable as the person (the same reasoning as Japan's Ministry of Justice guideline)

Labeling AI outputs (Measures for Labeling AI-Generated Synthetic Content, effective September 1, 2025)¶

AI-generated or synthesized text, images, audio, video, and virtual scenes must carry both a visible label (e.g., "AI-generated") and an invisible label such as metadata
Visible labels are required for chatbots, AI writing, synthetic voice, face generation or swapping, and generating or editing immersive scenes
Where download or export is offered, files must contain the label
Malicious removal, alteration, forgery, or concealment of labels is prohibited

Anthropomorphic AI interaction services (Interim Measures, effective July 15, 2026)¶

Ban on "virtual partner/family" services for minors
Guardian consent for users under 14
Mandatory identification of minors and a minor mode (time limits, spending limits, guardian controls)
Clear indication that it is AI
A usage-time pop-up after more than 2 hours of continuous use
Warnings when signs of excessive dependence are detected
App stores must check security assessments and not list non-compliant apps

Implications for entertainment businesses¶

💡 The world's strictest regulation of AI characters and companions. For global releases, making a minor mode and the 2-hour notice standard features makes other regions easier to handle
Labeling can be satisfied together with EU Art. 50 and Korea's AI Basic Act through a common design (visible labels + machine-readable marks)

Sources (checked 2026-09-24 to 25)¶

Bird & Bird (anthropomorphic AI) https://www.twobirds.com/en/insights/2026/china/china's-new-regulations-on-ai-anthropomorphic-interactive-services
Loeb & Loeb (labeling measures) https://www.loeb.com/en/insights/publications/2025/03/chinas-ai-labeling-measures-and-mandatory-national-standards-take-effect-september-1
China Law Translate (English translation of the labeling measures) https://www.chinalawtranslate.com/en/ai-labeling/
China Law Translate (English translation of the generative AI interim measures) https://www.chinalawtranslate.com/en/generative-ai-interim/
King & Wood Mallesons (first AI voice ruling) https://www.kwm.com/cn/en/insights/latest-thinking/china-s-first-case-regarding-ai-generated-voice-infringement.html