软餐监控的数据显示，成千上万个由 AI 生成的 “虚拟人” 短视频正在蜂拥涌向 YouTube、Facebook、TikTok、Instagram、X（Twitter）等主流社交媒体平台。我们重点追踪的几百个账号正在日夜不停的利用 AI 创建视频内容，这些内容相当一部分来自中国。为了快速获取流量，这些账号的操纵者从搜索引擎上批量获取色情内容，并由 AI 虚拟人以自述的方式阅读并生成视频。我们发现某些视频的播放量已突破千万。由于这些视频画面中的虚拟人声情并茂，不仔细甄别很容易误以为是真人。大量不明真相的观众对这些视频发表了评论。
社交媒体不会完全阻止虚拟人视频，但可能会要求创作者向观众声明视频主角是虚拟人，并且仍会沿用当前的内容审核政策来审核这些视频。
Thousands of AI-generated “virtual human” short videos are flooding towards mainstream social media platforms such as YouTube, Facebook, TikTok, Instagram, and X (Twitter). We have been closely monitoring hundreds of accounts that continuously use AI to create video content, with a significant portion originating from China. In order to quickly gain traffic, the operators of these accounts disclose obtaining pornographic content from search engines and generate videos in a self-narrative manner through AI virtual humans. We have found that the views on certain videos have already surpassed tens of millions. Due to the vivid portrayal of emotions by virtual humans in these videos, it is easy for uninformed viewers to mistake them for real people without careful discernment. A large number of unsuspecting audiences have commented on these videos.
Social media platforms may not completely block virtual human videos but may require creators to declare to the audience that the protagonist in the video is a virtual human. They will continue using their current content moderation policies to review these videos.
微信扫一扫