Video game streaming giant Twitch has introduced a setting that allows users to opt out of Amazon using their channel content to train generative AI models, more than two years after a company executive publicly confirmed the practice.
Ars Technica reports that Twitch announced the AI opt out option on Wednesday through an updated support page. The streaming platform, which Amazon acquired in 2014, confirmed that user content may be used to train what it calls Amazon’s “generative AI content models” unless users take action to opt out.
The support page specifies the types of content covered by the policy. Users who do not opt out agree that their “streams, VODs, clips, stream chats, and pictures and text on your channel [may] be used in future training of a model developed by Amazon whose purpose is to generate or synthesize text, audio, images, or video.”
How to Opt Out
- Navigate to the security section of Twitch settings by clicking here. You may be required to sign in to your account.
- Scroll down to “Training for Generative AI” and disable the feature.
- Confirm the change by navigating to another section of setting and returning to security. Your screen should look like the screenshot below.
Twitch’s support page explains what permitting the training entails: “When you allow training, that means that your content may be used for future Gen AI model improvements. An example of what happens when you allow your content to be used for training a Gen AI content model is that your 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.”
Twitch CEO Dan Clancy has faced critcism from users for not spreading the announcement via his own platform.
The new opt-out option arrives well after Amazon’s use of Twitch content for AI training became public knowledge. Twitch users are automatically enrolled in the training program and appear to have been for at least a couple of years. At a 2024 event hosted by the news publication The Information, Twitch’s chief monetization officer at the time, Mike Minton, was asked whether Amazon uses Twitch to train AI models. He replied, “Yeah, for sure.” He added: “I mean, I think obviously within the bounds of user trust within the bounds of privacy regulations, which vary all around the world, but of course we have a role to play in that.”
The structure of the new control matters to how it will function in practice. Because enrollment is automatic, every Twitch user who does not visit the security settings page and change the default will continue to have their content available for training. Opt-in-by-default arrangements tend to preserve the status quo, since many users never review or adjust account settings.
That design, combined with the length of time the training appears to have occurred without a clear way to decline it, may leave some users doubtful about Amazon’s approach to user-generated content. The company now offers a formal mechanism for withholding consent going forward, but the policy does not address content already used to train models before the opt-out became available.
Read more at Ars Technica here.
Lucas Nolan is a reporter for Breitbart News covering issues of AI, free speech, and online censorship.