Yep, the last part in my post was suggesting some ways we could determine if "item X was in the training data" (as well as some potential blockers for that from

Yep, the last part in my post was suggesting some ways we could determine if "item X was in the training data" (as well as some potential blockers for that from OpenAI's perspective.)