OpenAI have admitted their new model they used was trained on prompts at around the time that researcher was working on it, so it seems self evident that it was used as part of the millennium solution