I think putting up pinned post on the bug report tracker might help with this. People can clearly see that bug reports generated by AI are not allowed. The post can also explain what uses of AI are and are not allowed. For example, you can use AI to do things so long as you verify that what the AI is doing is correct.

Won't stop everyone but at the very least I think it will reduce the people with good intentions doing it.

The hallucination thing is often characterised as an unfortunate side effect or exceptional, whereas it is really inevitable from how these models are designed, particularly how they score their own output.

We can imagine marketing a chess program that produces moves quickly and "wins" because it doesn't check for legal moves, or one invents a fairy piece to tackle a position (while yes fairy chess is a real thing where both sides agree on new piece types with special moves).