Back IT之家科技新闻

IT之家科技新闻· 2026-10-28 16:56 · Original publication time AI score26

Doubao spokesperson denies Hengdian performers’ data will train ByteDance modelsMachine translation

豆包公关负责人否认横店群演数据用于训练字节大模型

Automatically verified and published · Generated and evidence-checked automatically; not reviewed by a human.

AI-assisted summary

Doubao spokesperson Liu Xing said a rumor that Hengdian and ByteDance would use the images and voices of 150,000 performers to train ByteDance models was seriously false. He said the Hengdian film and TV AI service platform aims to help production teams scout locations, prepare and shoot using assets such as film-set locations. Use of anyone’s image or voice would require their authorization, and project data would not be used to train ByteDance models.

Article · Original

IT之家 10 月 28 日消息，豆包公关负责人刘星今日发文称，“横店与字节合作，将 15 万名群演的肖像和声音喂给字节大模型”的传言严重失实。

他表示，横店影视 AI 服务平台项目，其实是希望建设一套公共数字基础设施，把影视基地实景等资产做成高质量数据集，帮助剧组更高效地选景、筹备和拍摄，也降低中小影视企业使用 AI 的门槛，让更多创作者有条件把作品做好。

IT之家注意到，文中提到，火山引擎在该项目中是技术支持方，所有数据资产，都归横店或授权方所有。这个项目涉及到任何人的肖像和声音使用，肯定都需要获得本人的授权。本项目的数据也不会用于训练字节大模型。