Once you're at the 12th month of trying to shoehorn LLMs in several use cases at your job, you'll find the answer to this question:

BECAUSE YOU CAN'T FUCKING TRUST THOSE LYING HALLUCINATING PIECES OF SHIT.

Clearly you just set an LLM to respond to messages that appear to be written by LLMs, then disregard that thread from that point on.

It's the same problem, false positives.

And false negatives too.