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

Twitch announced an update to its account settings that allows streamers to opt out of having their content used to train Amazon's artificial intelligence models. The change can be made from the Twitch website or mobile app by navigating to Settings, then Security and Privacy, where users will find a new toggle labeled 'Generative AI Training.' When disabled, this toggle prevents Amazon from using the creator's streams, posts, and videos for AI model training.

However, the update comes with an important caveat. Twitch clarifies that disabling the Generative AI Training option "does not prevent Twitch and Amazon from using your channel's content for other purposes described in Twitch's Privacy Notice." These other purposes include AI-powered platform features designed to help creators grow and monetize their channels, such as real-time assistance with sponsorship campaigns, viewer discovery through recommendations, and community safety tools like AutoMod.

The revelation that Twitch and Amazon had been using creator content for AI training without explicit opt-out mechanisms sparked significant backlash. More than 16,000 creators expressed opposition in a forum dedicated to the topic, according to the article. The controversy intensified after Mary Kish, Twitch's head of community, and Mike Minton, Twitch's head of product, addressed the issue in a livestream. Kish acknowledged that the change would provoke negative reaction, while Minton described his response as "candid," noting that keeping the AI training option enabled by default was necessary because "no one would participate" if it required active opt-in. Minton also observed that Twitch may not be alone in this practice, saying: "I don't know for sure, but I think it's quite reasonable to assume that almost any publicly available content is used to train models in one way or another, with or without permission."

The timing and scope of Twitch's use of creator content for AI training remain unclear. Twitch's Terms of Service, which took effect in March 2024, state that users grant Twitch and its sublicensees the right to use, reproduce, modify, adapt, distribute, and create derivative works from their content. However, until the recent update, the terms did not explicitly mention generative AI model training. The article notes that this case highlights a broader industry challenge: as AI developers compete to build more capable models, the supply of high-quality training data has become scarce. Major companies including Meta, Google, and others have similarly turned to user-generated content from their platforms—including posts, images, employee activity, and screenshots from smart-glasses devices—to fuel their AI training pipelines. In this ecosystem, free digital services are increasingly exchanged for access to users' creative and behavioral data.