Twitch, the live streaming platform, recently announced that it will start using creators' live stream content to help train generative AI models for its parent company Amazon. This move has sparked strong resistance within the Twitch creator community, with the focus being on the policy's "opt-in by default" mechanism, which means streamers need to manually opt out without being aware in order to avoid their audio and video content being used to train AI.

During a live event on the official channel, Twitch Community Manager Mary Kish and Chief Product Officer Mike Minton faced questions from nearly 3,000 viewers. In response to the overwhelming number of messages asking "Why not make it opt-out by default?" Minton bluntly replied that if it were set as "opt-out by default," no one would be willing to participate voluntarily.

image.png

Twitch management admitted that the platform is well aware of the general opposition among streamers toward generative AI, as most AI products are trained using publicly available data from the internet without the creators' consent. Because of this, when announcing this news, the platform deliberately framed it as an "optional setting to opt out of Amazon's company-wide AI model training," which caused many streamers to worry and get confused about whether their content had already been secretly collected.

Kish added that allowing users to opt out was meant to align with the community's firm rejection of helping train AI. Similar to other tech giants that directly use public platform data for AI training, although Twitch provides an opt-out option, this move once again highlights the deep conflict between the platform and creators over AI copyright and data usage rights.

For streamers who wish to refuse participation, they can currently find the "Generative AI Training" option by scrolling down to the "Safety & Privacy" tab in their channel settings (not the Creator Dashboard) and manually turn it off.