What does it mean when an AI model hallucinates?

AI hallucination happens when a model generates information that sounds convincing but is completely made up — like confidently telling you that the Eiffel Tower is in London. It's not lying in the human sense. The model has no concept of truth. It's simply producing text that follows the patterns it learned during training, and sometimes those patterns lead to plausible-sounding fiction. A well-known example happened in 2023 when a lawyer used ChatGPT to prepare a court filing, and the AI invented six fake legal cases with fake citations. The lawyer didn't verify them, and it became a major embarrassment. That's the danger of hallucination — the output looks exactly as credible as real information. Why does this happen? LLMs are trained to predict likely word sequences, not to verify facts. When the training data doesn't contain the answer, the model doesn't say "I don't know." It fills the gap with the most statistically probable words, which often sound right. I've found that hallucinations are most common when you ask about obscure topics, recent events after the model's training cutoff, or very specific numbers like statistics and dates. The practical tip: always verify AI-generated facts before using them in anything that matters. Ask the model to cite sources, then actually check those sources. For numbers, cross-reference with a search engine. And if you're using AI for research, treat it as a starting point for finding leads, not as the final authority. For a deeper dive, see our guide on [AI content copyright and legal issues](/blog/ai-content-copyright-legal-issues). **Related**: How can I prevent AI hallucinations? | Do all AI models hallucinate equally?