Twitch has offered creators a switch to keep their work out of Amazon’s future generative-AI training. The catch is the one that has inflamed its community: creators are enrolled unless they act.

The controversy has roots in Amazon’s long ownership of Twitch, which it acquired in 2014, and in a disclosure that predated the new control. At a 2024 event, then-chief monetization officer Mike Minton answered “Yeah, for sure” when asked whether Amazon used Twitch content to train AI, while invoking privacy rules and user trust.¹

This week, Twitch formalized an opt-out setting covering streams, VODs, clips, chat and channel images and text for future Amazon models that generate or synthesize text, audio, images or video. The company says such material could improve tools including speech-to-text captions on Twitch and across Amazon.²

But the rollout also confirmed why many streamers see the setting less as consent than as cleanup. On Twitch’s explanatory livestream, head of community Mary Kish acknowledged the reaction was unlikely to be warm: “We don't expect you to be happy or excited about this.”³ Live chat filled with anti-AI comments, while prominent creators urged users to disable the setting, worried their voices and hours of broadcasts could be absorbed without clear compensation or meaningful notice.³

Minton’s defense sharpened that divide. Asked why participation was not voluntary from the outset, he said: “If this was opt-in, nobody would opt in. That’s honestly the answer.”⁴ He and Kish argue Twitch is reflecting a broader internet reality, where public material is widely used in model training, and presenting an opt-out as a response to community pressure. Critics see the same logic as evidence that Amazon’s need for data outweighed creators’ right to affirmatively agree.⁴

The practical remedy is straightforward—users can turn off “training for generative AI” in Twitch’s security and privacy settings—but the central question remains unresolved: how much Twitch content Amazon may already have used. Minton said he did not know what Amazon had trained on, leaving the new safeguard to address the future while uncertainty hangs over the past.³