Glossary
What is AI hallucination?
An AI hallucination is when a model states something false or invented with confidence — including wrong facts about your brand or a competitor that doesn't fit. Monitoring AI answers catches hallucinations about you early, so you can correct the underlying sources before they spread.
Why hallucinations matter for brands
An assistant can describe your business with a discontinued product, a feature you never built, the wrong city, or the details of a namesake company. Buyers may act on that without ever checking your site, and you only find out if you ask. The risk is not only being missing from answers but being present and wrong.
Why they happen and how to respond
Language models generate plausible text. Where information is missing, outdated or contradictory, they may fill the gap with something that sounds right. Grounding on retrieved sources reduces this but does not remove it.
A workable process: ask the engines factual questions about your brand on a schedule; log the prompt, the engine, the date and the answer; trace the claim to a source by checking the cited pages and searching for the wrong statement; fix it at the source, whether that is your own page or an outdated listing; publish clear facts in an about, product or FAQ page; and test again after a few weeks. You cannot edit a model directly.
A hypothetical example
For example, a bakery in Lima finds that an assistant says it delivers nationwide and is closed on Sundays. Neither is true. Searching for those claims turns up an old directory listing and a blog post from two years earlier. The owner corrects the listing, asks the blog to update its post and adds a delivery and opening-hours page to her own site. She then asks the same questions again over the following weeks.
Common mistakes
- Overreacting to one odd answer. Responses vary between runs, so confirm the pattern first.
- Arguing with the chatbot. Correcting it in a conversation does not change what it tells other users.
- Ignoring the source of the error, which is usually a page you or a third party can fix.
- Not keeping a record, which makes it impossible to tell whether things improved.
Common questions
Can I ask an AI company to correct a wrong statement about my brand?
Most assistants offer feedback controls, and some companies provide channels for privacy or legal requests. Results vary, so fixing the underlying sources is usually more reliable.
How do I find hallucinations about my brand?
Ask each engine factual questions about your products, location, features and policies, then compare the answers with reality. Repeat over time, because a claim that appears once may not be a pattern.
See if AI assistants recommend your brand
Run a free scan to see which AI engines mention your brand and who they cite instead of you.
Scan my site free