China Supreme Court sets liability for unauthorized AI deepfakes and voices

The new legal framework forces streaming platforms to implement more aggressive content moderation for synthetic media or risk direct liability for user-generated infringements. By targeting AI-driven celebrity impersonations in live-stream sales, the court is addressing a specific friction point in China's massive social commerce market where deepfakes have undermined consumer trust. This move signals a shift toward proactive judicial intervention rather than slow, post-facto litigation, potentially setting a precedent for how other jurisdictions handle the speed of AI-generated harm. Watch for whether these injunctions lead to the development of automated deepfake detection mandates for all major Chinese video service providers. For more details on the specific requirements, see the China Supreme People's Court issues 24-article China AI deepfake guidance.

China's Supreme People's Court guidance arrives amid a broader wave of regulatory and judicial actions targeting synthetic media across the Asia-Pacific region. In South Korea, the National Assembly passed amendments to the Information and Communications Network Act in June 2026 that impose criminal penalties of up to five years imprisonment for distributing non-consensual deepfake content, a significant escalation from the civil remedies previously available. Japan's Ministry of Internal Affairs and Communications similarly published draft guidelines in May 2026 requiring video-sharing platforms operating in Japan to label AI-generated content and remove unauthorized synthetic likenesses within 24 hours of a verified complaint. These parallel moves suggest that China's judicial approach, which relies on court-issued injunctions rather than statutory mandates, represents one model among several that regional regulators are testing to address the same underlying problem of unauthorized synthetic media proliferation on streaming and social platforms.

The business implications extend well beyond content moderation budgets. Wang Liming, a former dean of Renmin University Law School who advised the Supreme People's Court on the new opinion, argued in a July 2026 legal commentary that the framework deliberately places platform intermediaries in a gatekeeper role analogous to the notice-and-takedown regimes that shaped early internet copyright law. For streaming and live-commerce operators, this means that failure to act on a court injunction could trigger direct liability rather than the safe-harbor protections that previously shielded platforms from user-generated infringement. The Supreme People's Court has not yet published enforcement statistics, but legal analysts at the China Academy of Information and Communications Technology estimated in an August 2026 report that Chinese short-video and live-streaming platforms received more than 12,000 deepfake-related complaints in the first half of 2026 alone, a volume that underscores the operational scale of the compliance challenge.

On the technical side, detection tooling is racing to keep pace with the regulatory timeline. China's Ministry of Industry and Information Technology certified three domestic deepfake detection systems in April 2026 under its AI content authentication pilot program, including tools from SenseTime and iFlytek that can identify synthetic voice and facial artifacts in real-time video streams. The certification is notable because it creates a de facto technical standard that platforms can reference when demonstrating compliance with the court's injunction requirements. Meanwhile, , a effort that could eventually harmonize the labeling and detection approaches that China, South Korea, and Japan are each pursuing independently. For streaming platforms operating across multiple Asian markets, the convergence or divergence of these technical standards will determine whether compliance requires separate detection stacks per jurisdiction or a single unified system.