Twitch, owned by Amazon, has implemented a default opt-in system for AI training that automatically feeds live streams, video archives, and chat logs into Amazon's generative AI training. According to reports from The Verge, the switch sits at the very bottom of the Security and Privacy tab, making it nearly impossible for users to find without a walkthrough. The company framed its August 12 announcement as a new privacy control, but most streamers interpreted the wording as proof that the collection had already started. Creators have recently spotted a "Training for Generative AI" option in account security that's toggled on by default, with the description reading "Allow your channel content to train generative AI content models at Amazon."
Twitch Support has told users that the company itself does not train generative models on streamer content, with its own help page still listing gen AI model improvements among the uses, which points to Amazon. According to The Verge, Chief Product Officer Mike Minton defended the default during a livestream on Wednesday, stating that if it was opt-in, nobody would opt in. "Well, there's an honest answer," Minton said, per Kotaku. "And I think most of you probably can appreciate this: If it was opt-in, nobody would opt in. Um, that's honestly the answer. So it's going to be on by default." During the stream, Minton referenced a recent study claiming that 50 percent of adults are now using AI tools like ChatGPT, and confirmed he was "comfortable" with AI training, stating he would not opt out if he was a streamer. Many creators fear their voices and faces will be used to train digital clones that undercut them later, similar to ongoing legal battles where authors are suing Anthropic over book piracy and Reddit is weighing whether to cut Google's AI data access.
Twitch states that the material can sharpen speech-to-text captions and other generative systems across Amazon. As reported by The Verge, broadcasts, saved videos, and comments all qualify for the AI training program. According to the latest reports, every part of a Twitch channel can be used for training Amazon AI models, including streams, VODs, clips, chats, pictures, and text on your channel. A community note on the announcement flagged a second catch: opting out will not shield a viewer who types in someone else's chat, because that channel's own setting decides. This gap pushed anger well past the creator base and into the wider Twitch user pool. According to Twitch's FAQs, "When you allow training, that means that your content may be used for future Gen AI model improvements." An example provided shows how audio might help refine models that create speech-to-text, which would help improve captions at Twitch but would also help improve captions across Amazon. Some users have also questioned the implications for game developers, given that the feature could allow Amazon's AI models to be trained using the large number of video games played by streamers.
The option to opt out of having your Twitch content used by Amazon to train generative AI is somewhat buried in the Settings. Users can find the "Training for Generative AI" toggle by clicking on their profile picture, then heading to Settings, and then Security and Privacy. As reported by Mashable, toggling off won't affect AI-supported features on your channel, such as captions and safety tools. During the livestream, Twitch's Head of Community Mary Kish walked viewers through how to access the setting, seemingly admitting it is a bit hidden. However, some users have reported that the setting appeared to turn back on after they had disabled it. The backlash has been significant, with users expressing anger over the default setting and the lack of clear communication about the program's scope and implications. Head of community Mary Kish acknowledged the community's concerns, stating that "The opt-out numbers will help us continue to reiterate that our community is not receptive of AI in any form."
Twitch has confirmed it has no plans to implement an option for users to easily see which channels are opted in and opted out of AI training, according to the latest reports. When asked why Twitch didn't email users about the change, Kish said it was a "good question" and claimed that "not everyone reads their emails" and that DMs and word-of-mouth get the information "further." During the stream, Minton seemed not to have an answer when asked what would happen if streamers used someone else's channel that hadn't opted out, admitting it was a "good question." The company's Head of Community Mary Kish also claimed that "Twitch can't function without some forms of AI," while confirming she has personally opted out. It isn't surprising that Amazon, desperate to win the "AI WAR," would start tapping into Twitch to help train its generative AI models, as many companies are desperate for more content to feed to their AI machines as they compete with ChatGPT and Google. Minton also said he did not know whether users' data had already been scraped for AI training, stating he was unsure what Amazon "has done in terms of model training and what they've used and not used."
