Twitch enrolls streamers in Amazon AI training by default

Twitch announced that it will use creators' channel content to help train generative AI models for its parent company, Amazon, and that participation is opt-out rather than opt-in: unless a streamer manually turns the setting off, their content is included by default. Twitch framed the announcement narrowly, telling users it was 'adding a setting that lets you opt out of having your channel content used to train generative AI content models across Amazon' rather than stating outright that Amazon would begin training on their material. For Amazon, Twitch's stream recordings are valuable training data: thousands of hours of audio and video from creators, many of whom record livestreams of themselves and their voices for hours every week.

The backlash was swift. Twitch Head of Community Mary Kish and Chief Product Officer Mike Minton addressed the controversy live on the official Twitch channel, in front of a live audience of nearly 3,000 users, many of whom filled the chat with anti-AI messages. Asked directly why the setting was not opt-in, Minton did not dispute the premise: 'Why is it not opt-in? That's what everybody is spamming in chat. I get it. "Let me opt in versus making me opt out." Well, there's an honest answer... If this was opt-in, nobody would opt in. That's honestly the answer.'

The framing also produced confusion about content already collected. When a user asked whether their existing videos had already been used to train Amazon's models, Minton said he did not know: 'I don't actually know the answer to that question because I don't know what Amazon [...] has done in terms of model training and what they've used and not used.' The article does not say how, or whether, previously collected content would be affected if a creator opts out later.

Kish argued Twitch is not unique in training on user content, pointing to Meta's use of public Facebook and Instagram posts to train its own AI models: if a creator's accounts are public, their data has likely already been used. She noted that U.K. users can opt out of Meta's training, but elsewhere the only way to opt out is to make an account private, which is not workable for creators who rely on public accounts to monetize their channels. Kish described even offering an opt-out on Twitch as 'a reflection of reacting to this community's voice that you are not wanting to train Gen AI models, and we want to give you that option.'

To opt out, Twitch streamers go to channel settings (not the creator dashboard), open the security and privacy tab, scroll to 'training for generative AI,' and toggle it off.

Key facts

Twitch will train Amazon's generative AI models on creators' channel content by default; streamers must manually opt out rather than opt in.
Twitch executives Mary Kish (Head of Community) and Mike Minton (Chief Product Officer) addressed the backlash live, in front of a live audience of nearly 3,000 users, many posting anti-AI messages.
Minton admitted the opt-out design was deliberate: 'If this was opt-in, nobody would opt in.'
Minton could not say whether creators' existing content had already been used for training before the setting existed.
Kish compared the move to Meta, which trains on public Facebook and Instagram content; U.K. users can opt out, but elsewhere the only alternative is going private, which is impractical for creators who monetize public accounts.

Why it matters

This is a default-in design choice, and Twitch's own product chief said so on camera: an opt-in setting would have drawn far fewer streamers into AI training, so Twitch chose opt-out instead. It is a clear, publicly confirmed example of a platform using default settings to secure user data for AI training against the stated wishes of most of that user base, in an industry where the pattern of scraping creative work without explicit consent is already a live grievance.

Who it affects

Every Twitch streamer whose channel content had not already been excluded is opted in automatically. The impact is heaviest for creators who livestream themselves and their voices for many hours each week, since that is exactly the audio and video Amazon can use to train models. Streamers who rely on public accounts to earn money are also squeezed by the comparison Twitch itself drew to Meta: going private to avoid training is not a realistic option for anyone monetizing a public channel.

How to use it

To opt out, go to channel settings (not the creator dashboard) on Twitch, open the security and privacy tab, scroll to 'training for generative AI,' and toggle it off. The article does not specify a date for when the setting appeared or a deadline for opting out before content is used.

How solid is it

The account comes directly from Twitch's own leadership speaking on the record in a public livestream watched by nearly 3,000 people, including on-camera admissions about the opt-out design and about not knowing whether prior content had already been used for training. That is a strong, primary-source basis for the facts reported here, though the article itself does not name which specific AI models or products the content will train, nor give any figures for how many streamers have since opted out.

Risks and caveats

Because participation is default-on, creators who do not notice the setting, or do not know where to find it, are training Amazon's AI models without having made an active choice. Twitch has not said, at least according to this account, what happens to content already collected if a streamer opts out afterward, and the framing Twitch used publicly (announcing an 'opt-out setting' rather than announcing that training would begin) appears designed to soften how the change reads.

“Why is it not opt-in? That's what everybody is spamming in chat. I get it. "Let me opt in versus making me opt out." Well, there's an honest answer... If this was opt-in, nobody would opt in. That's honestly the answer.”

— Mike Minton, Twitch Chief Product Officer