Twitch opts streamers into Amazon AI training by default, sparking backlash over the setting

Twitch announced on August 12 that streamers can now opt out of having their content used to train Amazon's generative AI models. The catch: the setting is enabled by default, meaning Amazon has been using — and will continue using — a streamer's broadcasts, clips, chat, and channel content for AI training unless that person actively finds the toggle and turns it off.

The new control lives under Security and Privacy in account settings, labeled "Training for Generative AI." Turning it off stops Amazon from using a channel's live streams, videos on demand, clips, highlights, chat logs, and images in future model training. It does not undo any training that has already happened on past content.

Why this matters beyond Twitch

Twitch is owned by Amazon, and this is the clearest public acknowledgment yet that the platform has been quietly feeding creator content into Amazon's AI training pipeline. Livestreaming content is a genuinely valuable and rare training resource — unscripted speech, real-time reactions, natural conversation, and visual variety that's expensive to source any other way. For millions of streamers who spent years building an audience and a personal on-camera style, discovering that footage may already be embedded in a commercial AI model — without having been asked — is the core of the backlash.

The opt-out defense

During a livestream addressing user questions, Twitch chief product officer Mike Minton was asked directly why the setting wasn't opt-in. His answer was blunt: "If this was opt-in, nobody would opt in. That's honestly the answer." That's an unusually candid admission — Minton effectively confirmed that Twitch's own leadership expects creators wouldn't consent to this use of their content if given a genuine choice, and built the settings page around that expectation instead of around it.

The response from the streaming community has been sharp. A support-forum comment demanding the setting be flipped to opt-in has drawn close to 14,000 upvotes — an unusually large and fast mobilization for a platform-settings complaint.

Part of a bigger pattern

Twitch's move follows a now-familiar sequence across the industry: platforms quietly use user-generated content for AI training, get caught or forced to disclose it, then ship an opt-out toggle framed as a privacy improvement rather than address the underlying consent question. Similar controversies have played out with stock photo libraries, writing platforms, and social networks over the past two years. What distinguishes the Twitch case is the product chief's direct, on-record admission that the default was chosen specifically because the alternative would produce too little training data — a rare moment where a company said the quiet part about opt-out defaults out loud.

For streamers who want to check or change their setting, it's located in Twitch's account privacy settings under the AI training toggle. As reported by TechCrunch.