Twitch to use streamer content for Amazon AI training, sparking user criticism
Twitch announced on Wednesday that it will use content streamed on its platform to train Amazon's generative artificial intelligence models. This decision has drawn significant criticism from many users, particularly because the option to opt-out of this data usage is enabled by default. Twitch stated on X that it has added a setting allowing users to refuse the use of their channel content for training Amazon's generative AI models. However, users must actively disable this option if they do not wish their content to be used. The content in question includes live streams, video-on-demand, clips, chats, images, and text from channels. Twitch noted that audio recordings, for example, could help improve speech-to-text transcription models. The platform did not specify if content had already been used for Amazon's AI models or when this new policy would officially begin. Mike Minton, a product manager at Twitch, acknowledged the community's frustration regarding the default activation, stating, "This feature will be activated by default, as is the case on almost all content services worldwide." He added, "I know this upsets many in the community, but that's how it is." Twitch also clarified that opting out of generative AI model training does not exclude users from all AI uses, as some AI-based features, such as automatic subtitles, will continue to function. Amazon is not the first tech giant to use user data to train generative AI models; Meta Facebook, Instagram previously an