EVENT DOSSIER
Amazon has implemented a policy that automatically uses Twitch content to train AI; streamers must actively opt out to protect their content.
As of August 2026, Amazon announced that it would default to using content streamed by streamers on Twitch to train its artificial intelligence models. According to TechCrunch, unless a streamer explicitly chooses to opt out, their live streaming content will be included in the training data. Wired further noted that this mechanism aims to allow AI to learn how users can exit such services, but in practice, consentless content may still be used for model optimization. This strategy has already been implemented, and relevant streamers must take active measures to protect their content from being used.
回溯归档across 2 days
Amazon has implemented a policy that automatically uses Twitch content to train AI; streamers must actively opt out to protect their content.
Amazon announced that it will automatically use streamer content to train generative AI, unless creators manually opt out; Twitch responded that if the setting was set to “agree,” no one would agree, so the default mode was changed.
Twitch announced that it will use user viewing data (mouse trajectories, clicks, comments) to optimize AI, allowing users to turn off this feature in their settings to exit the training process.
亚马逊将默认使用 Twitch 主播内容训练其生成式 AI 模型，除非创作者手动选择退出。Twitch 社区对此迅速反弹，担忧主播在不知情的情况下让渡内容。Twitch 首席产品官 Mike Minton 在直播中回应称，若设为“同意”模式无人会同意，因此改为默认使用并允许退出。部分主播因混淆已发生的数据使用情况而质疑内容是否已被训练。Twitch 社区普遍反对生成式 AI，且 Meta 等平台对公开内容的 AI 训练做法引发类似争议。目前用户需进入频道设置的安全与隐私选项，选择“用于训练”下的退出功能以阻止内容被使用。
亚马逊利用 Twitch 用户内容训练其人工智能，Twitch 已发布相关说明。该服务通过收集用户观看视频时的数据（包括鼠标移动轨迹、点击行为及评论）来优化 AI 算法。Twitch 表示这些数据用于改进推荐系统和广告定位，并允许用户在设置中关闭数据收集功能以退出训练流程。