Twitch is now staring down the barrel of a copyright lawsuit related to artificial intelligence. The core grievance is rather simple: the platform allegedly harvested streamers' content to train its AI models without so much as a polite 'by your leave.'

Back in August, Twitch streamer Warren Pandisia filed a class-action lawsuit against Amazon and Twitch in the U.S. District Court for the Northern District of California. The complaint alleges that the corporations used massive quantities of broadcast content to train generative artificial intelligence models without securing consent or offering fair compensation. Pandisia argues this constitutes a breach of the implicit agreement between Twitch and creators, as well as a violation of state unfair competition laws.

To rewind a bit, Twitch announced in early August that it would begin training generative AI models using creator content—including broadcasts, past clips, chat logs, and photos generated by streamers within the platform. Right now, Twitch automatically ropes user content into AI training unless individuals manually opt out. These settings apply on a channel-by-channel basis, meaning that even if you choose to opt out, your content might still end up in the AI training grinder if you participate in a channel that has consented to AI usage.

Naturally, the announcement went down like a lead balloon in the streamer community. Creators immediately pushed back, questioning why on earth this feature was enabled by default, forcing users to actively jump through hoops to opt out of AI training.

Amidst all this friction, Pandisia claims that Amazon and Twitch have secretly been using streamer content for AI training since 2024, completely bypassing any public disclosure or opt-out mechanism. So far, Amazon and Twitch have remained tight-lipped, issuing no official statement on the matter.