AI hallucination occurs when an LLM generates plausible-sounding output that is factually incorrect or internally inconsistent — stated with full confidence