# AI hallucination

> Learn why AI systems fabricate convincing but false details and how to protect your business with reliable, machine-readable sources.

Synonyms: hallucination, AI hallucinations, fabricated AI answers

Hallucinations stem from how large language models work: they compute which phrasing is most likely — not which one is documented. Where reliable information on a question is missing, the model fills the gap with something plausible. The result sounds confident but may be entirely made up: a wrong phone number, a closing day that never existed, a service you do not offer.

The most effective remedy from a business's point of view: give the AI reliable material. An up-to-date website as the central source of truth, consistent contact details everywhere and structured data that labels facts unambiguously. Modern AI systems consult real sources before answering — a business that presents itself clearly and machine-readably there gets reported correctly instead of guessed at.

For your business that means checking regularly what assistants say about you — opening hours, prices, services. If wrong details appear, the cause is almost always an outdated or contradictory source. The fix happens not inside the AI but at the source: bring your website, Google profile and directories to the same, current state.

---
Source: https://btlabs.dev/en/glossary/ai-hallucination
Last-Modified: 2026-07-21T07:00:00.307Z
Languages: [de](https://btlabs.dev/llms/de/glossary/ki-halluzination) · [it](https://btlabs.dev/llms/it/glossary/allucinazione-ia)
See also: [llms.txt](https://btlabs.dev/llms.txt) · [ai.txt (Policy)](https://btlabs.dev/ai.txt) · [identity.json](https://btlabs.dev/identity.json)