"training a large language model off my text" - we don't do that, and our privacy policy is very clear about how we use the text.