AI hallucinations are when an AI model generates information that sounds confident and plausible but is factually wrong. The model isn't lying — it doesn't understand truth. It predicts what text SHOULD come next based on patterns, and sometimes those patterns produce convincing-sounding nonsense. Common examples: citing research papers that don't exist, inventing statistics, describing events that never happened, or attributing quotes to people who never said them. Hallucinations are the single biggest risk in deploying AI for business use. The mitigation is: always verify AI outputs against primary sources, use RAG to ground the model in real data, and never publish AI-generated content without human review.