# Hallucination (AI)

> A hallucination is when an AI model produces false or non-existent information that looks true, such as a made-up quote, number or reference.

## Why it happens

LLMs are trained to produce plausible text, not to check facts. When information is missing, they fill the gap with what "sounds right".

## How to reduce it

- Provide sources in the prompt (RAG) and ask the answer to cite them.
- Let the model say "I don't know".
- Ask for verifiable answers: links, numbers with a source.
- Always review when a mistake has a cost.