GEO GLOSSARY
A confident but factually incorrect statement produced by an AI model — for example inventing a feature, price or fact about a brand that the model was never actually given evidence for.
A hallucinated detail about pricing, availability or a feature can create real friction when a prospective customer arrives expecting something that was never true. This is a brand-risk problem, not only a visibility one — a wrong but favorable-sounding claim is not a win.
No public-facing content change can guarantee a model never hallucinates, but keeping factual pages current, unambiguous and easy to retrieve reduces how often a model has to fill gaps with guesses. Ongoing monitoring — checking what is actually being said — catches problems that one-time optimization cannot prevent.
See how these terms are used in practice in MagRank AI’s methodology.
