Summary
China’s new judicial guidance on AI face and voice synthesis has been in effect for a month as courts and voice performers press for stronger provenance. The guidance says using a person’s voice without consent to train a system or generate an identifiable imitation may violate voice rights. On September 29, a Shanghai court concluded the city’s first dispute over AI-synthesized speech, finding a platform operator liable and awarding the performer 50,000 yuan despite the company saying a former employee’s training inputs could not be traced. The issue also reached animation: director Sun Chenxi said a voice generated through an external model for the AI animated film “Nie Xiaoqian” closely resembled actor Yang Tianxiang. The disputed video was removed and replaced with human dubbing. Voice studio 729 Sound Factory is calling for consent-based voice collection, clearer licensing trails and stronger warnings from generation and distribution platforms.
Hana’s perspective
I think this is an important correction to a dangerous assumption: paying for an AI tool does not prove that its training material or generated voice is licensed. For animation and microdrama teams, voice provenance now needs to be documented as carefully as music, footage and character rights. The Shanghai ruling also shifts attention away from a model’s internal route and toward whether the output identifies a real person without consent. That makes “we could not trace the data” a production failure, not a defense. Platforms should expose licensing records, but producers still need approval checks before release.
Are you a writer with a great story? We'd love to hear from you: APPLY HERE