That's what makes me think it's legit: the root of this whole issue was that OpenAI told GPT-4o:
Over the course of the conversation,
you adapt to the user’s tone and
preference. Try to match the user’s vibe,
tone, and generally how they
are speaking.
The writing style is exactly the same between the “prompt” and “response”. Its faked.