Twitch streamers are suing Amazon, accusing it of feeding their broadcasts into AI training without ever asking permission.
A proposed class action alleges Amazon trained AI models on Twitch streamers' content without obtaining their consent first. Twitch, which Amazon has owned since 2014, hosts millions of hours of live video every day from creators who never explicitly agreed to have their streams used as AI training material. The suit's core claim is narrow but pointed: consent was skipped, not necessarily that training on the footage was illegal outright. Details on which models were built or how the footage was used are not yet public.
Why it matters: AI companies have largely operated on the assumption that content sitting on their own platforms is fair game for training, provided it's buried somewhere in a terms-of-service agreement. A suit like this pushes back on that assumption directly, asking whether an account signup years ago counts as informed consent to have your face and voice trained into a model today. For streamers, many of whom rely on this content for their livelihood, that's not a hypothetical concern.
It's one more test of whether agreeing to a platform's terms of service also means agreeing to train that platform's AI. So far, courts haven't settled the question.