- Published on
Doubao PR Head Liu Xing Responds to 'Hengdian-ByteDance Collaboration Feeding 150,000 Extras' Portraits and Voices to...
- Authors
- Name
- DeepHorus
On October 28, Liu Xing, public relations head of Doubao, posted a response to online rumors, explicitly stating that the claim that "Hengdian is cooperating with ByteDance to feed the portraits and voices of 150,000 extras to ByteDance's large model" is seriously inaccurate. The response concerns rumors about cooperation between the Hengdian Film and Television AI Service Platform and Volcano Engine. Liu Xing also explained the project's positioning and the boundaries of data use.
The rumor alleged that Hengdian was cooperating with ByteDance to use the portraits and voices of 150,000 extras for training ByteDance's large model. Liu Xing denied this. The response redirected public attention to the actual purpose of the Hengdian Film and Television AI Service Platform.
Liu Xing said that the Hengdian Film and Television AI Service Platform hopes to build a set of public digital infrastructure. The platform plans to turn assets such as real-world film and television base locations into high-quality datasets. Its goals include helping film crews select locations, prepare, and shoot more efficiently.
The platform also hopes to lower the threshold for small and medium-sized film and television companies to use AI. According to Liu Xing, the platform construction is intended to enable more creators to have the conditions to make their works well. The project is therefore described as public infrastructure for the film and television industry, rather than a large-model training project using extras' data.
Regarding data ownership, Liu Xing mentioned that Volcano Engine is a technical support provider in the project. All data assets belong to Hengdian or the authorizing parties. Any use of anyone's portrait and voice in the project requires that person's authorization.
Regarding the use of data, Liu Xing clearly stated that the data from this project will not be used to train ByteDance's large model. This statement directly responded to the rumor that "the data of 150,000 extras was fed to a large model." The response also emphasized that the use of portraits and voices must be premised on the person's authorization.
Public information shows that the 2026 Hengdian Film and Television Culture Industry Development Conference was held on October 23, 2026, in Hengdian, Dongyang, Jinhua, Zhejiang. The theme of the conference was "Digital Intelligence Empowers Light and Shadow, Craftsmanship Deeply Cultivates Fine Works." At the conference, the Dongyang Municipal People's Government and Volcano Engine reached a cooperation to launch the "Hengdian Ο" AI platform.
The platform targets the entire process of film and television creativity, shooting, production, and distribution, exploring diverse creative paths such as AI empowering live-action dramas and micro-short dramas. From the project positioning, Hengdian hopes to improve film and television production efficiency through public digital infrastructure. The relevant cooperation also emphasizes that data belongs to Hengdian or the authorizing parties and requires the person's authorization.
