Twitch Defaults Creator Content Into Amazon AI Training

Twitch added a default-enabled setting allowing channel material to train Amazon’s generative AI models, with creators required to opt out manually. Eligible content includes streams, VODs, clips, highlights, text, images, and chat. Twitch says resulting models could serve products across Amazon, including speech-to-text and captioning. Chief product officer Mike Minton said the setting was not opt-in because nobody would choose to participate.

Twitch has enabled Amazon generative-AI training on users’ channel material unless they opt out. Its FAQ says covered data includes streams and chats, VODs, clips, highlights, text, and images. The article provides instructions for disabling the setting under Security and Privacy and notes that opting out does not disable AI-supported features such as captioning, AutoMod, or other safety tools. It emphasizes widespread anger over the default.

Twitch will use creators’ content to train Amazon generative AI models by default, generating swift opposition from streamers. The article says Amazon gains extensive audio and video material while creators may unknowingly surrender recordings of their voices and appearances unless they manually opt out. During a stream with nearly 3,000 aggrieved viewers, Mike Minton acknowledged that Twitch chose an opt-out structure because nobody would affirmatively participate.

Twitch reportedly activated a default-on setting on August 12 covering streams, archived videos, clips, chat logs, and channel images for Amazon AI training. The article says creators received neither advance notice nor a clear account of material previously ingested during prototyping dating to 2024. It portrays Mike Minton’s explanation—that nobody would opt in—as evidence Twitch deliberately chose a default that bypassed users’ expected refusal.

Twitch announced that Amazon plans to collect platform material for AI training and then faced intense hostility during a question-and-answer stream. Mike Minton said the system is opt-out because nobody would affirmatively join, prompting viewers to argue that Twitch should abandon it. The article says Minton and Mary Kish failed to justify the policy and describes their appeal to similar data collection by OpenAI and Anthropic as an inadequate defense.

Twitch says streams, VODs, comments, and other creator material may train generative AI systems across Amazon unless creators opt out. The policy was indirectly confirmed through Twitch Support’s announcement of the disabling control and drew overwhelmingly hostile responses. Mike Minton said the system defaults to participation because nobody would opt in. A community proposal demanding optional, default-off AI features had reached 228 pages and nearly 14,000 supporting votes.

Twitch appears to have begun using livestreams, VODs, and other streamer content to train Amazon AI models, with a newly discovered opt-out buried in account security settings. Reporter Zach Bussey publicized the setting after it was added with little apparent warning. Disabling it stops generative-model training but does not block Twitch’s other AI-powered functions, including automated commenting, advertising selection, and chat moderation. Kotaku sought further clarification from Twitch.

Twitch introduced a default-enabled setting permitting Amazon to train generative AI models on users’ channel content. Users can disable it through the Security and Privacy tab, although doing so does not prevent Amazon from using channel material for other purposes under Twitch’s privacy notice. The article condemns the default and cites a Twitch forum request for optional, default-off AI features that has attracted more than 13,000 votes and 4,000 comments.

Twitch acknowledged using streamed material to train generative AI models for Amazon and introduced a deeply placed, default-enabled control requiring manual opt-out. The article reports that executive Mike Minton could not specify whether personal data had been captured or precisely what Amazon had already done in model training. It also says users received no login notification and portrays the policy, its implementation, and the executives’ explanations in sharply condemnatory terms.

Twitch now includes creators’ streams, archives, and chat logs in Amazon AI training by default while presenting the change as a privacy control. The opt-out sits at the bottom of Security and Privacy, and the article says viewers lack their own control when posting in another channel’s chat. Twitch says the material may improve speech-to-text captions and other Amazon systems, while executives defended the policy as responsive to community feedback.

Twitch will make creators’ material available for training Amazon generative AI models unless they manually opt out, prompting immediate community backlash. The article says recordings offer Amazon extensive audio and video training data but may include streamers’ voices and appearances without an affirmative choice. During an official stream attended by more than 3,000 viewers, Mary Kish and Mike Minton faced numerous anti-AI comments and questions about the announcement.

The article alleges Twitch quietly enrolled every creator’s past broadcasts, clips, chat messages, and VODs in Amazon AI training without explicit permission or compensation. It characterizes the opt-out as hidden deep in account settings and says users learned of it through reporters rather than direct notification. Citing Mike Minton’s statement that virtually nobody would opt in, it argues management intentionally prioritized acquiring creators’ work over affirmative consent.

Twitch faces international creator backlash after enabling Amazon to use livestreams, chat logs, and audio for generative-AI training by default. The article argues that placing the opt-out deep in privacy settings shifts responsibility for protecting data onto users and disproportionately exposes creators worldwide. During an official broadcast, Mike Minton said Twitch rejected an opt-in model because nobody would participate, an explanation the article presents as evidence of aggressive corporate data harvesting.