Twitch streamers can now opt out from training Amazon’s AI
Twitch has added a privacy control allowing creators to opt out of having their content used to train Amazon's generative AI models. The setting covers a broad range of material tied to a creator's presence on the platform - streams, VODs, clips, stream chats, and pictures and text published to their channel. Importantly, Twitch specifies that the opt-out applies to future training, meaning content that may have already been used in model training is not retroactively excluded.
The opt-out is scoped specifically to generative AI - systems whose purpose is to produce or synthesize text, audio, images, or video. Other AI-supported features that Twitch operates, such as automated captions and moderation and safety tools, will continue to function regardless of whether a creator opts out. This distinction matters because it separates the data-sharing question from the platform's day-to-day utility for creators.
One nuance worth noting involves chat participation. When a creator opts out, the preference governs their own channel's chat data, but if that same creator participates in chat on someone else's stream, it is the other streamer's opt-out preference that determines whether those messages can be used for training. This means a creator's chat contributions could still be captured as training data on channels where no opt-out has been set.
The move reflects growing pressure on platforms that host large volumes of user-generated content to give creators more explicit control over how that material feeds into AI development. Twitch sits within Amazon's broader ecosystem, which includes its own generative AI initiatives, making the provenance of training data a particularly pointed question for the streamer community. Offering an opt-out - rather than an opt-in - puts the default in Amazon's favor, but it does at least surface the choice for creators who seek it out.
