Twitch has confirmed that it uses creator content to train Amazon’s generative AI models, prompting the platform to introduce an opt-out setting for all streamers.

Key Points

• Twitch automatically enables AI training for all channels, requiring users to manually navigate to account settings to opt out.
• The collected data, including audio, video, and text, is used to improve Amazon’s generative AI models and speech-to-text capabilities.
• Users can disable the training feature while still retaining access to other platform tools like Automods and campaign optimization.
• Concerns remain regarding whether audience chat contributions are included in the training data and how rebroadcasting affects creator consent.
• The use of copyrighted music within streams poses potential legal risks for Amazon regarding how its AI models ingest and process third-party intellectual property.

Why it Matters

This development highlights the growing tension between platform data collection practices and the intellectual property rights of content creators. As major tech companies integrate user-generated content into AI development, the burden of managing privacy and consent increasingly shifts to the individual user.