Amazon Can Use Your Twitch Content to Train Its AI—Unless You Opt Out

Details

Twitch quietly updated its privacy settings to let users opt out of having their content used to train Amazon's AI models, bringing an undisclosed practice into the open. The new toggle lives in Settings → Security and Privacy under 'Generative AI Training.' Critically, disabling it does not prevent Twitch or Amazon from using content for other AI-powered features—AutoMod, sponsorship tools, and viewer recommendations are all unaffected. Only model training use is blocked. The backlash was swift. More than 16,000 creators flooded a dedicated forum opposing the default opt-in. Twitch's head of product Mike Minton acknowledged the choice was deliberate: keeping it enabled by default was necessary because 'otherwise no one would participate.' He also suggested that virtually all publicly available content is likely being used to train AI models industry-wide—with or without explicit permission. Open questions remain: When exactly did Amazon begin using Twitch content for training? Are business partners also accessing this data? The timing of past usage is undisclosed, and Twitch's Terms of Service history is unaddressed in the reporting. The opt-out framing—rather than opt-in consent—fits a broader pattern in how major platforms handle AI training data. With growing EU and US legislative interest in AI training data transparency, this disclosure may invite regulatory scrutiny.