Twitch introduces opt-out for Amazon generative AI training

Twitch has introduced a new opt-out mechanism allowing users to prevent their content from being used to train Amazon’s generative AI models. The update provides a specific toggle within the platform’s settings, giving streamers greater control over how their digital assets are utilised in artificial intelligence development.

The “Training for Generative AI” toggle is located in the Security and Privacy tab of Twitch settings. According to the platform’s support page, enabling this option ensures that streams, video on demand (VOD) recordings, clips, stream chats, and channel pictures or text are excluded from future training of Amazon AI models designed to generate or synthesize text, audio, images, or video.

Opting out does not disable other AI-supported functionalities on the platform. Features such as captions, safety tools like AutoMod, viewer recommendations, and real-time sponsorship assistance will continue to operate. The privacy notice clarifies that while generative training is halted, Twitch and Amazon may still use channel content for other purposes aimed at facilitating streamer growth, monetisation, and community safety.

The scope of the opt-out extends to interactions on other channels. If a user participates in the chat of another streamer, that host’s opt-out preferences determine whether the chat content can be used for training. This creates a dependency on the content creator’s settings rather than the viewer’s, ensuring that the host’s data preferences govern the usage of the conversation.

The default status of the new toggle remains unconfirmed pending official verification from Amazon. While initial observations by reporters suggest the setting may be toggled on by default, this has not been formally validated. The move comes amidst broader industry scrutiny regarding the use of user-generated content for AI training, a debate intensified by recent controversies involving AI-generated media in promotional campaigns.