It makes no sense to me that one would train a chatbot on chatgpt conversations and not filter strings that literally say "openai" and "chatgpt". Extreme incompetence.
Handling the huge number of cases where ChatGPT says something like "As a large language model created by OpenAI" would be very simple.