Twitch and parent company Amazon are facing a class action lawsuit alleging that streamers’ broadcasts, clips, chat records and photos were used without permission or compensation to train Amazon’s generative AI models. Warren Pandiscia, a Connecticut streamer with about 1,000 followers, filed a 37-page complaint in the U.S. District Court for the Northern District of California, claiming Twitch breached an implied agreement and violated California’s Unfair Competition Law. The lawsuit is also the first public confirmation that Amazon is using Twitch content to train its own AI models, although it remains unclear when the data collection began or how much content has been used. The dispute follows Twitch’s August 13, 2026, announcement of an opt-out setting for channel content, which indicated that AI training was enabled by default, and criticism of Twitch Chief Product Officer Mike Minton’s comment that nobody would opt in if training were voluntary. Amazon acquired Twitch for nearly $1 billion in 2014 as it has increased AI investment to compete with companies including Google and Meta. Twitch content is also embedded in prediction-market platforms such as Polymarket, which has more than 500 markets that rely on Twitch livestreams of esports events, broadening the potential significance of the case for creator rights and data licensing.