So grok is specifically weird because it's can respond directly to requests. I would imagine OpenAI could put a warning into their Chat that says, replies are attributable to you and not meant to be shared publicly to avoid liability, but grok maybe not so much.

Is Elon saying publicly that he changed the prompt to include "politically incorrect responses" prove intent?