I kind of already assumed they were. I've got some pretty niche use-cases that I'd like to see the models get better at thinking their way through. I benefit from their training on my interactions. So I'll opt in. But I'll also recognize that others might not feel that way, so the services should provide a way for users to opt out.

Ditto: was delighted to see this as an option. Am I missing some details? I do not understand the comments in “rug pulls”.