Shanghai court fines AI app $105,000 for Genshin voice clones
A Shanghai court ordered an AI voice‑changing app to stop selling voice packs based on 63 Genshin Impact characters and to pay HoYoverse RMB 750,000 (about $105,000).
The Shanghai Pudong New Area People’s Court found an AI voice‑changing app used audio tied to 63 Genshin Impact characters without permission and ruled the practice amounted to unfair competition. The operator was ordered to stop offering the character voice resources and to pay HoYoverse RMB 750,000 (about $105,000) in damages.
HoYoverse discovered in 2025 that the app was selling paid voice packs and character images marketed as reproductions of the game’s voices. The app operator told the court the sounds were produced by an AI model and argued that rights belonged to the individual voice actors.
A voiceprint assessment submitted in the case concluded the generated voices were likely to be identified as the same as the original character voices. During proceedings the operator acknowledged using about one minute of audio from Genshin Impact character recordings, mixed with other voice samples, to train its AI model without HoYoverse’s authorization.
The court found that the training method, the marketed voice packs and the use of character images created confusion about the source of the voices and harmed HoYoverse’s commercial interests. The ruling requires the operator to stop offering the unauthorized character voice resources and to pay the ordered damages.
Court documents describe the decision as Shanghai’s first unfair competition judgment involving AI voice imitation. The ruling follows recent guidance from China’s Supreme People’s Court on handling AI‑related disputes, including cases where synthetic voices are created without consent and can be linked to an identifiable original speaker.
HoYoverse, the developer of Genshin Impact, has previously pursued legal action to protect its intellectual property. The company filed the lawsuit to challenge the use of copyrighted character voices and images in AI training and in paid voice products.







