Twitch streamers were surprised this week to discover that their streams and content were being used to train Amazon AI models, according to a setting buried in its menus. It’s not just that the setting was implemented, but that it’s turned on by default for every user, and you have to manually opt-out if you don’t want to be a part of that.
This led many users to decry the practice, saying that it should be opt-in instead, rather than forcing them to extract themselves manually. This escalated to the point where on a livestream with Twitch CPO Mike Minton, he gave a shockingly plain answer that may be obvious, but it’s pretty wild to hear it said out loud:
“Now why is it not opt-in?” said Minton. “That’s what everybody here is spamming in chat. I get it. ‘Let me opt in versus making me opt out.’ There’s an honest answer, and I think most of you can probably appreciate this: If it was opt-in, nobody would opt in. That’s honestly the answer.”
The stream, which also featured head of community Mary Kish, talked about how auto opt-in is standard industry practice. Minton also said that other places from OpenAI to Anthropic are likely already scraping things like public Twitch streams without any sort of consent. There also seems to be some daylight between Amazon and the Twitch team here.
Kish: “The opt-out numbers will help us continue to reiterate that our community is not receptive of AI in any form. I’m sorry, but I am reading your comments and do hear you.”
Minton: “Twitch has not been training models. I can’t speak to what Amazon is training or not training as it relates to any specific usage of content.”
This entire situation is, of course, a reflection of global AI training, where billions of pieces of content are taken and used for training without permission, at times resulting in lawsuits, though there is no sign of this kind of practice slowing down any time soon. Google, for instance, justifies it like this:
“Using publicly available web data for training models is a transformative, non-expressive use — like an art student taking inspiration from walking through a gallery — that should remain protected under fair use in the U.S. and text-and-data-mining exceptions abroad.”
It may not be a surprise that Twitch’s response did not placate fans, but rather made the reaction even more intense. Users who have been made aware of this are now rushing to disable the feature. And here’s how to do that:
- Click on your profile picture in the top right
- Go to Settings
- Go to Security and Privacy
- Scroll pretty far down and uncheck the box for Training for Generative AI
There are some caveats, however. Namely, if you chat in someone else’s stream, changing this does not protect that, given that it relies on that person’s settings, not your own.
We will see if Twitch has anything else to say about this, but given that Amazon is calling the shots here, it’s unlikely anything of significance will change. Though Minton is right, anything you post publicly online is likely being scraped by dozens of AI models, they’re just not telling you.
I’ve asked Twitch for further comment and will update if I hear back.
Follow me on Twitter, YouTube, and Instagram.
Pick up my sci-fi novels the Herokiller series and The Earthborn Trilogy.