Overview
- A proposed class-action complaint filed on August 20 alleges Twitch and Amazon copied stream video, clips, chat logs and images to train commercial generative AI and that scraping began as early as 2024.
- Twitch disclosed on August 12 that it uses creator content for Amazon's AI and made 'Training for Generative AI' a default-on setting that users must opt out of in their channel settings.
- The suit says the opt-out is nonretroactive and tied to individual channels so a creator who disables the setting can still have their content captured when they appear on a channel that remains opted in.
- Pandiscia seeks injunctive relief, damages, restitution and disgorgement and argues the harm is effectively permanent because training data cannot be 'clawed back' once incorporated into models.
- Twitch and Amazon have not publicly responded to requests for comment and the case could shape future legal rules on consent, licensing and remedies for AI training data use.