Twitch has added a settings option allowing streamers to stop their content from training Amazon’s artificial intelligence systems.
It is not clear exactly when the streaming platform started using posts, streams, and videos to train Amazon’s models. The update has sparked fresh worries about how big tech firms manage user data.
The setting is straightforward to change. Go to your account avatar on the Twitch website or app, select Settings, then find the Security and Privacy section. There, you will see the Generative AI Training option to disable.
Next to the toggle, Twitch states that turning this off does not stop Twitch and Amazon from using your channel for other purposes in their Privacy Notice. These include AI features for streamer growth and monetisation, like real-time sponsorship help, viewer discovery recommendations, and community safety tools such as AutoMod.
The new setting recognises creators’ right to decide how their output is used. Its introduction has also prompted questions about how Twitch and Amazon handled that content up to this point.
In a dedicated forum, more than 16,000 creators expressed opposition to having their content used by default to train Amazon’s AI. This practice only became visible after the settings update.
The backlash followed a livestream where Mary Kish, Twitch’s head of community, explained the changes. The executive admitted the move would provoke a negative reaction. Mike Minton, Twitch’s head of product, also noted in what he called a candid response that enabling the option by default was necessary, otherwise no one would participate.
Minton pointed out these mechanisms are not unique to Twitch. He considered it likely other companies developing AI systems are also extracting content from Twitch and other services. “I don’t know for sure,” he said, “but I think it’s quite reasonable to assume that almost any publicly available content is used to train models in one way or another, with or without permission. So I think we also need to acknowledge that there’s a lot here that’s beyond even our direct control.”
These statements raised new questions. Since when has Twitch content been used to train AI models? Is Amazon the only company using this data, or are its business partners involved too? To what extent and in what ways is the authorship of creator content respected?
Twitch’s Terms of Service, in effect since March 2024, state users grant Twitch and its sublicensees the right to use, reproduce, modify, adapt, distribute, and create derivative works from their content. Until now, they did not explicitly say such materials could be used to train generative AI models.
The Training Data Problem
This case highlights a major challenge for AI developers: a growing shortage of high-quality data to train models.
Although companies like OpenAI have reached agreements with various publishers, including Condé Nast, to use some content for this purpose, available data sources are dwindling as technology advances and demand increases. As a result, various alternatives have emerged, reigniting debate over the ethics and transparency of training processes.
For some time, Meta has used posts and images shared on Facebook and Instagram to train its AI models. Recently, it also came to light the company was using employee activity and screenshots captured by users of its smart glasses for the same purpose. Similar cases have been documented involving Google and YouTube.
High-quality training data has become another strategic resource, alongside chips, memory, and electricity, that leading AI developers have found running short. They struggle to sustain the pace of innovation the market demands. All signs point to a portion of that demand being met with information generated by users themselves, in exchange for free access to everyday digital services.
This story originally appeared in WIRED en Español and has been translated from Spanish.
