Twitch Streamers Can Now Opt Out of Amazon AI Training Data

Twitch has introduced a new privacy control allowing creators to block Amazon from using their streams and videos to train artificial intelligence systems. The opt-out feature came as a surprise to many—not because the option itself is controversial, but because the practice had apparently been underway without explicit disclosure.

How to Disable AI Training on Twitch#

The process is straightforward. Streamers can navigate to their account avatar on Twitch's website or app, select Settings, then open Security and Privacy. Within that menu lies a toggle for Generative AI Training that can be switched off to prevent content from being used for AI model development.

However, the fine print matters: Twitch clarifies that disabling this setting doesn't stop the company from deploying creator content for other AI-powered features. These include sponsorship recommendations, viewer discovery algorithms, and community moderation tools like AutoMod.

The Backlash and Unanswered Questions#

The announcement sparked significant outcry. Over 16,000 creators voiced opposition in Twitch's community forums, objecting to the default opt-in arrangement. The controversy deepened when executives offered candid explanations: Mary Kish, Twitch's head of community, acknowledged the change would provoke negative reactions, while Mike Minton, head of product, argued that flipping the default to opt-out "would mean no one would participate" in AI training.

Minton also raised a broader point—that Amazon likely isn't alone in scraping publicly available Twitch content for AI purposes, and that most companies developing large language models probably extract data from multiple sources regardless of explicit permission.

The disclosures left crucial questions unanswered: When exactly did Twitch begin using content for AI training? Is Amazon the sole beneficiary, or do business partners also access this data? How are creator rights actually protected?

A Symptom of the Data Crisis#

This situation reflects a mounting challenge in AI development: companies are running short of high-quality training material. While OpenAI and others have negotiated licensing deals with publishers, the demand for data continues to outpace available sources. As a result, tech giants are increasingly turning to user-generated content as an alternative.