Twitch has drawn sharp criticism from its user base after revealing that content generated by streamers and viewers is used by default to train Amazon's artificial intelligence models. The Amazon-owned streaming platform confirmed that a setting permitting the collection of streams, video-on-demand, clips, chat logs, and channel images for generative AI training is enabled automatically, though users can disable it in their privacy settings.
Chief Product Officer Mike Minton acknowledged the rationale during a livestream, stating that making the feature opt-in would yield virtually no participants. Community head Mary Kish walked viewers through the opt-out process but downplayed concerns, noting that not all users read email notifications about policy changes. A user petition demanding the setting be reversed to opt-in has garnered more than 13,000 supportive votes.
Twitch stated that opting out does not affect other AI-driven features such as recommendations and moderation tools. The controversy echoes broader industry debates, with YouTube having launched an opt-in AI training feature in late 2024 and Meta using European users' public content for model training since May 2025.