Back to Glossary
TechnicalAI Glossary

Hallucination

Quick Answer

A hallucination is when an AI model produces output that sounds plausible but is factually incorrect, fabricated or inconsistent with its sources. Hallucinations are a fundamental property of current generative models and the single biggest risk in enterprise deployments.

In Depth

What Hallucination really means

Hallucinations arise because language models are optimised to generate likely continuations, not truthful ones. They can invent citations, misremember numbers, blend similar facts, or assert incorrect causal relationships with high confidence.

Mitigations include retrieval augmentation, grounding prompts with authoritative sources, requesting citations, constrained output formats, and human-in-the-loop review for high-stakes outputs.

Why It Matters

Business relevance for UK organisations

For UK businesses in regulated sectors — financial services, healthcare, legal, insurance — unmanaged hallucinations can cause compliance breaches and reputational harm. Building guardrails into every customer-facing AI product is non-negotiable.

Real-world example

How this shows up in practice

A London wealth manager narrowly avoided sending a client an AI-drafted letter that cited a non-existent FCA regulation, after a human reviewer caught the fabricated reference.