I mean, they don’t train on your data unless you have the setting enabled. Do you really think they are reading your prompts at all? Free inference providers sure, but Anthropic?