The longer story
An AI hallucination is when a model generates information that sounds confident and plausible but is factually wrong or entirely made up. This happens because language models predict what text should come next based on patterns, not because they know facts. Hallucinations can range from minor inaccuracies to completely fabricated citations, statistics, or events. This is why human review of AI outputs is essential, especially for anything factual, legal, or medical.