LLM Hallucination – Why AI Makes Things Up, How to Detect It, and Which Models Hallucinate Most

LLM hallucination, or confabulation per NIST, ranges 1% to 60%+ by task. Courts have documented 120+ fake AI citation cases.