Every "on by default, opt out if you notice" AI policy is dodging the same question, who actually owns the value inside the data a model touches. @Twitch answered it this week, turning training on for every streamer's account, livestreams, VODs, clips and chat logs included, unless you dig up the setting yourself. Its own chief product officer admitted why, saying plainly that if this were opt-in, nobody would opt in. Real ownership has to be a default the system enforces on its own. Sahara built the Data Services Platform around that idea, putting consent into the infrastructure itself rather than a settings page someone has to remember to check.(Sahara AI 🔆)