A hallucination is when an AI generates information that is confident-sounding but false or unsupported by any real source. Because language models predict plausible text rather than looking facts up, they can state incorrect dates, invent details, or misdescribe a brand without any signal that the answer is wrong.
For brands, hallucinations are a reputation risk: an AI might describe your product inaccurately or attribute the wrong facts to you. Grounding answers in retrieved sources reduces hallucination, which is one reason a clear, accurate, well-structured web presence matters. When the correct facts about you are easy to retrieve, the AI is less likely to fill the gap with something wrong.