This is the fundamental problem and shortsightedness of the ChatGPT ban on Stack Overflow - it doesn't go far enough.

There is some imagined distinction between a hallucinated answer from a LLM and one from a person posting "try {random code}" that are both equally wrong - but only the LLM is deleted (because deleting the wrong answers are an attempt by a human to answer(?) or would hurt feelings / engagement (?)).