Hallucination
When an AI tool states something false or made up as if it were true, such as a fake quote, a wrong date or a book that does not exist.
A large language model writes by predicting likely words, not by looking facts up in a reliable database. Most of the time this produces sensible answers. Sometimes it produces text that sounds right but is not, and it does so in the same confident tone.
A common example: you ask a chatbot for sources on a topic and it lists articles with plausible titles and authors, but some of them were never written. Lawyers, students and journalists have all been caught out by invented references.
The word "hallucination" can make it sound like a rare glitch. It is a normal risk of how these systems work, and newer tools reduce it without removing it. Be most careful with names, numbers, dates, quotes, medical, legal and financial details. Check them against a trusted source before you rely on them.
Related: Large language model (LLM), Chatbot, Training data