Twitch users: Amazon could use your content for AI training. Here's what you should know

Twitch, the live streaming service owned by Amazon, has introduced a default setting that allows the use of streamer content to train Amazon's artificial intelligence models. This has generated significant criticism from the Twitch community, as users must manually opt out of this data usage rather than explicitly agreeing to it.

Amazon sees Twitch as a rich source of video and audio content for AI model training, given the extensive time many streamers dedicate to their platforms. Twitch initially framed the update as an opt-out feature, which has led to confusion about whether any content had already been used for AI training. During a live stream, the company's Chief Product Officer, Mike Minton, confirmed that creators had the option to refuse content use but admitted to not knowing what, if any, data Amazon had previously used.

Minton defended the decision, stating that if creators were required to opt-in, few would do so. He also noted that publicly available content might already be scraped by other AI companies, whether permitted or not. While opting out will not affect a creator's discoverability on Twitch, it only prevents the usage of their content for generative AI training, not other AI or machine learning features like AutoMod, captions, and platform moderation.

To opt out, users can disable the setting in either the Twitch web interface or mobile app, with the preference applying across both platforms.