Twitch has automatically enrolled its creator community in an opt-out data-harvesting policy that allows parent company Amazon to scrape livestreams, chat logs, and VODs for generative artificial intelligence training. The quiet policy rollout sparked intense creator backlash across livestream broadcasts and social media platforms.
The streaming platform Twitch announced a policy change that lets users opt out of Amazon’s use of content from their channels to train its generative artificial intelligence content models. Amazon, which purchased Twitch for nearly $1bn (£740m) in 2014, has been scaling its AI infrastructure rapidly as it competes with major technology companies like Google and Meta.
The Minton Moment: Why Twitch Chose an Opt-Out Default
The rollout quickly generated controversy because Twitch users were automatically opted into the AI training by default. During an official platform broadcast, Twitch Chief Product Officer Mike Minton and Head of Community Mary Kish addressed an audience of nearly 3,000 users while the live chat filled with anti-AI messages. When questioned about why the platform relied on an opt-out mechanism rather than requiring affirmative consent, Minton offered a candid explanation.
“If this was opt-in, nobody would opt in. That’s honestly the answer.”
Mike Minton, Chief Product Officer at Twitch
Kish acknowledged the negative reception during the broadcast, stating that leadership did not expect a favorable reaction from the community.
“We don’t expect you to be happy or excited about this. I don’t expect anyone to react to this favorably.”
Mary Kish, Head of Community at Twitch
What Creator Data Amazon’s AI Models Are Harvesting
The scope of content vulnerable to automated scraping is extensive. Unless a creator manually disables the feature, Amazon is authorized to ingest nearly every digital artifact associated with a channel. The harvested data pool includes high-definition video feeds, facial expressions, real-time audio capturing conversational nuances, community chat histories, archived Video on Demand (VOD) files, and short-form clips.
Twitch stated that training data helps improve specific platform features, such as using audio to refine speech-to-text models that enhance captions on Twitch and across Amazon. However, streamers expressed deep concern over whether their historical broadcasts had already been utilized. When pressed on whether Amazon had already absorbed past content prior to the announcement, Minton offered a sobering concession during the stream.
“I don’t know what Amazon in this case has done in terms of model training, and what they’ve used and not used. I’m not responsible for Amazon’s training efforts.”
Mike Minton, Chief Product Officer at Twitch
Creator Backlash and the Fight Over Consent
High-profile streamers and advocacy groups immediately criticized both the training practices and the opt-out architecture. Writing on X, Twitch user Legundo urged his followers, EVERYONE make sure to turn this off and tell Twitch we don’t want this crap!
Prominent partners such as HasanAbi, Pokimane, and Asmongold also voiced sharp criticism over the default data harvesting.
Streamers argued that their original performances should not serve as unpaid training datasets for one of the world’s most valuable corporations, especially when the resulting AI tools could eventually replicate their voices and creative styles.
How to Opt Out of AI Training on Twitch
For creators wishing to block their channel content from future AI ingestion, Twitch requires navigating away from the standard creator dashboard. Users must access their account settings via the proper settings at www.twitch.tv/settings/security, select the security and privacy tab, locate the toggle for training for generative AI near the bottom of the page, and switch it off.
Some users reported technical friction during the process. Twitch has stated that creators who choose to opt out will not face algorithmic penalties or reduced promotional visibility, and that private data will be filtered from training sets.
