There are lots of ways to detect AI hallucination, new guardian agent model promises to correct them.