A very simple example would be a mandatory mechanism for correcting mistakes in prebaked LLM outputs, and an ability to opt out of things like Gemini AI Overview on pages about you. Regulation isn't all or nothing, viewing it like that is reductive.