Amazon is training its internal artificial intelligence models on Twitch livestreams, sparking immediate backlash from creators and users on the streaming platform, according to reports published on Wednesday. Twitch, which was acquired by Amazon in 2014, announced the data usage alongside a new opt-out setting, officially confirming that creator content has been utilized for model training.
Twitch Confirms AI Training and Rolls Out Opt-Out Setting
During a community livestream addressing the platform updates, Twitch head of community Mary Kish stated that officials expected negative reactions from the user base. According to Business Insider, Kish told viewers, “We don’t expect you to be happy or excited about this,” adding, “I don’t expect anyone to react to this favorably.” The live chat accompanying the broadcast filled up with hundreds of anti-AI messages as creators voiced their displeasure.
To address privacy concerns, Twitch introduced a new control allowing users to opt out of having their content used for AI training. Users can access the setting by navigating to their privacy settings page and adjusting a slider at the bottom. However, because the setting defaults to an active state, users are automatically opted in unless they manually change the preference.
Creator Backlash and Leadership Responses
Several prominent streamers took to social media platform X to condemn both Amazon’s utilization of live content and the default opt-out mechanism. Twitch user Legundo posted on X, urging others to change their configurations, writing, “EVERYONE make sure to turn this off and tell Twitch we don’t want this crap!” according to Business Insider.
During the explanatory broadcast, Twitch chief product officer Mike Minton addressed questions regarding the timeline and scope of the data collection. Minton stated that he did not know whether Amazon had already been using Twitch livestreams to train models prior to the announcement. “I don’t know what Amazon in this case has done in terms of model training, and what they’ve used and not used,” Minton said, according to Business Insider. “I’m not responsible for Amazon’s training efforts.”
Minton and Kish defended the platform’s data practices by framing the opt-out mechanism as a unique privacy feature. Minton noted that “almost any publicly available content is being used for model training in one form or another, whether permitted or not.” Furthermore, Minton clarified that Twitch utilizes artificial intelligence strictly for platform utilities such as generating captions, clips, and content classifications, rather than running generative media features.
Amazon Scales Infrastructure for Frontier Model Race
The controversy arrives as parent company Amazon scales up its artificial intelligence infrastructure. In recent months, Amazon has wound down several internal models to position itself to compete directly in the frontier model race against competitors such as OpenAI and Anthropic. Representatives for Amazon did not immediately respond to requests for comment regarding the livestream data usage.