# Hallucination

> A confident model output that is false, unsupported, or not grounded in the available evidence.

- Category: Evaluation & Safety
- Stage: Established
- Also called: Fabrication, Unsupported answer
- Canonical: https://www.joelcsaji.com/ai-dictionary/hallucination

## Plain English

The AI makes something up and sounds sure about it.

## Example

A model invents a nonexistent API method while explaining a library.

## Why it matters

Hallucinations are a core reason teams use grounding, citations, evals, and human review.

## Related

- [Grounding](https://www.joelcsaji.com/ai-dictionary/grounding)
- [RAG](https://www.joelcsaji.com/ai-dictionary/retrieval-augmented-generation)
- [Evals](https://www.joelcsaji.com/ai-dictionary/evals)

## Sources

- [Google Cloud Generative AI Glossary](https://docs.cloud.google.com/docs/generative-ai/glossary)
- [IBM: Retrieval-Augmented Generation](https://www.ibm.com/think/topics/retrieval-augmented-generation)