Twitch Defaults Users Into Amazon AI Training
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Twitch sparked user backlash after it emerged the platform allows Amazon to use creator and audience data to train AI models, with the setting enabled by default and an opt-out process described as disrespectful by critics.
- Twitch CPO Mike Minton admitted the default-on design was deliberate, saying of an opt-in alternative: "If it's opt-in, nobody would opt-in. That's the honest answer" — and acknowledged he did not know whether user data had already been scraped for training.
- Twitch's head of community Mary Kish walked users through opting out via Settings → Security and Privacy → toggle off "training for Generative AI," though multiple users reported the switch reverting to "on" after they disabled it.
- Twitch's AI training could use channel audio to refine speech-to-text models for automatic subtitles on Twitch streams and Amazon videos, and would also seemingly train on video games being played by streamers — raising questions for game developers.
- Twitch said collected data will not be re-sold to other companies, and Kish called data collection for AI training "industry standard," while adding: "We don't expect you to be happy or excited about this."
- Amazon bought Twitch for nearly $1bn (£740m) in 2014 and runs multiple AI services as it competes with Google and Meta, though it is unclear when Amazon began collecting Twitch user data for AI training.
Why it matters: Twitch creators risk losing control over their streams, audio, and game footage to Amazon's AI training pipeline, with an opt-out that users report reverts to 'on.' Twitch's CPO couldn't confirm whether data has already been scraped, while framing the default-on policy as 'industry standard' — a template other platforms may copy.
Ask SkimNews
