Definition·
Limits

Hallucination

An AI response that sounds credible but is factually wrong or fabricated.

Detailed explanation

LLMs generate likely token sequences, not verified facts: they can invent sources, citations, dates or numbers. RAG, tools, human review and grounding on trusted sources reduce this risk.

Examples

Inventing nonexistent case law
Citing a book with a fake ISBN
Giving a wrong event date

Frequently asked questions

How to limit hallucinations?

RAG on internal sources, mandatory citations, human review, and preferring 'I don't know' over made-up answers.

Related terms

Last updated: 7/15/2026

Talent AI

Turn theory into practice

Post a mission or join the community of top AI, Data and Machine Learning experts.