Twitch enables default opt-out for Amazon AI training on streamer content

Summary

The BBC segment reports that Amazon-owned Twitch is using streamers' clips, videos, chats, images and other channel content to train its generative AI models. The "Training for Generative AI" setting is enabled by default; users must manually opt out via account settings. Twitch executives addressed backlash in a livestream, with chief product officer Mike Minton stating that an opt-in system would see near-zero participation. The head of community described AI data collection as an industry standard, and the company confirmed data would not be sold to third parties. The start date of training remains unclear.

Sourcing drew from Twitch's livestream, support pages, and direct executive statements. The BBC contacted both Twitch and Amazon for comment. No independent experts or affected streamers were interviewed in the clip; coverage focused on the announcement, user outrage on social media, and Amazon's broader AI investments.

Editorial Assessment

The report is factually accurate and closely tracks primary statements from Twitch's August 12, 2026 livestream and support documentation. Viewers receive the core controversy—default inclusion versus explicit consent—but miss important context: Twitch's terms of service and privacy notice grant a broad, perpetual license to create derivative works and explicitly reference use for machine learning and generative AI, meaning the new toggle is an added control rather than a reversal of prior policy. Reports indicate Amazon has used Twitch data for AI since at least 2024. The framing centers user outrage and the candid executive quote, which could amplify perceptions of corporate overreach while underplaying that similar opt-out models exist across Meta, YouTube and other platforms. The piece leaves unclear that opting out does not prevent all uses of data for AI-supported features like recommendations or moderation.