Twitch adds opt-out for Amazon AI training amid backlash

Twitch has introduced a new account setting that lets streamers opt out of having their content used to train Amazon's generative AI models. The setting, located under Security and Privacy, is enabled by default, meaning users must manually disable it to prevent their streams, VODs, clips, chat messages, and channel images from being used in future AI training Source: The Verge. The move follows a backlash from creators who were unaware their content was being used for this purpose Source: BBC.

The default controversy

Twitch's decision to make the setting opt-out rather than opt-in has drawn sharp criticism. Chief Product Officer Mike Minton acknowledged the reasoning plainly: "If this was opt-in, nobody would opt in. That's honestly the answer" Source: TechCrunch. This admission has been described as a "dark pattern" by some observers Source: The Next Web. Head of Community Mary Kish added that Twitch does not expect users to react favorably, but framed the opt-out as a response to community concerns Source: BBC.

Scope and limitations

Opting out prevents Amazon from using a user's content for generative AI training, but it does not stop other AI-powered features like AutoMod, captions, or recommendations Source: Engadget. Additionally, if a user participates in another streamer's chat, that streamer's settings determine whether the chat is used for training Source: The Verge. Twitch has not clarified when it began using content for AI training, and Minton said he does not know if users' data has already been used Source: Gizmodo.

Industry context

Twitch's practice is not unique; other platforms like YouTube and Meta also train AI on user content, often without clear opt-out options Source: Engadget. Minton noted that "almost any publicly available content is being used for model training in one form or another, whether permitted or not" Source: Engadget. The situation highlights broader concerns about data consent in AI development.