Amazon will train on Twitch streamers' content by default, opt-out only

Bulletin — Specola Galactica, Telescope Log

Amazon-owned Twitch will begin training AI systems on streamers’ content by default, requiring users to opt out rather than opt in, TechCrunch AI reported on August 12.

The policy surfaced after user feedback prompted Twitch Chief Product Officer Mike Minton to address the matter directly on a livestream. Asked why the company had not structured the program as opt-in, Minton offered an unusually direct answer: “If this was opt-in, nobody would opt in. That’s honestly the answer.”

The remark stands as a rare on-record acknowledgment from a platform executive that a consent mechanism was designed around the assumption that voluntary participation would fail to generate sufficient data. TechCrunch frames the episode as a test case for consent standards across the AI supply chain, where platforms holding large volumes of user-generated content increasingly face decisions about default settings versus affirmative opt-in.

The report does not specify further details of the training program’s scope, what content categories are included, or the mechanics of the opt-out process itself. No additional statements from Amazon or Twitch beyond Minton’s livestream comments are cited in the available reporting.

The story is presented as ongoing, with TechCrunch AI as the sole cited source for both the policy disclosure and Minton’s remarks.