Twitch has introduced a setting allowing creators to opt out of having their content used to train Amazon's generative AI models, following backlash over the default opt-in policy.

Twitch has confirmed that Amazon is using content from the streaming platform, including streams, clips, and chats, to train its generative AI models. The data collection is currently enabled by default for all users. In response to community criticism, the company has introduced a new setting that allows creators to manually opt out of the training process.

Rationale for Default Collection

During a livestream addressing the update, Twitch Chief Product Officer Mike Minton acknowledged the community's frustration regarding the default opt-in status. When asked why the company did not make the feature opt-in, Minton stated that if it were, no one would choose to participate. He confirmed that while the collected data will not be resold to other companies, he was uncertain whether user content had already been scraped for training purposes prior to the announcement.

Community and Industry Context

The announcement prompted significant negative feedback from the Twitch community, with many users expressing concerns about their creative work being used to train AI models without explicit consent. Twitch officials noted that using user-generated content for AI training has become an industry standard, citing similar practices by companies like Meta. Twitch staff provided a tutorial for creators on how to navigate their channel settings to disable the data collection.