Twitch are scraping streams for AI training by default, and even their own execs donβt know what user data may have been swept up
Livestreaming giants Twitch have admitted using streamed content to train the generative AI models of parent company Amazon. While itβs not clear when the platform began feeding its usersβ work into Jeffy Bβs Planetkilling Machines, yesterdayβs announcement of an opt-out toggle β buried deep within Twitchβs security settings, and set to enable AI stream-gobbling by default β sparked outrage among streamers and viewers that such scraping was occurring at all.
Twitchβs chief product officer Mike Minton and head of community Mary Kish were wheeled out to explain the policy in a stream of their own, though itβs possible the only ones reassured by the discussion were the Amazon bots themselves. Although Minton was upfront about the modelsβ use of streamed video, he conceded that he didnβt know if usersβ personal data had also been caught up in the scraping net, and that he wasnβt actually sure what Amazon "has done in terms of model training." Awesome. Awesome.
