Hallucination (AI)

A hallucination is when an AI model produces information that is false, fabricated, or unsupported by any source — while presenting it fluently and with apparent confidence.

AI Hallucination in Practice

In *Moffatt v. Air Canada* (2024), a customer asked the airline’s website chatbot about bereavement fares. It told him he could buy a full-price ticket and claim the discount retroactively within 90 days; the airline’s actual policy didn’t allow after-travel requests. The British Columbia Civil Resolution Tribunal held Air Canada liable for negligent misrepresentation, rejected its argument that the chatbot was a separate entity responsible for itself, and ordered CA$812.02 in total — damages, interest, and fees ([Canada Commons](https://canadacommons.ca/artifacts/43557256/moffatt-v-air-canada/44456228/); [McCarthy Tétrault](https://www.mccarthy.ca/en/insights/blogs/techlex/moffatt-v-air-canada-misrepresentation-ai-chatbot)). The same failure at smaller scale: an assistant tells shoppers that Northloom offers 90-day returns when the real window is 30 days, and the support team inherits the argument.


How to Audit AI Hallucinations About Your Brand

For a brand, tracking means a periodic manual audit: ask major AI tools factual questions about your company — pricing, features, policies, leadership, availability — compare each answer to ground truth, and log errors. Repeat monthly and across several runs, since the same prompt can return different answers each time.


Why AI Hallucination Matters

An invented claim about your product can be repeated to many people at once, and the Air Canada ruling shows the party deploying a chatbot can be held accountable for what it says. For third-party AI, you can’t fix the model — but you can publish clear, consistent, crawlable first-party facts that a retrieval system can ground on, and Google says grounding is meant to reduce hallucinations ([Google AI for Developers](https://ai.google.dev/gemini-api/docs/generate-content/google-search)).


Terms Related to AI Hallucination

Grounding (AI) ·
Retrieval-Augmented Generation (RAG) ·
Parametric Memory
Knowledge Cutoff


FAQ

Can I stop AI tools from hallucinating about my brand?
Not directly — training data and model behavior are out of your hands. You can reduce the odds by keeping official facts consistent across your site and profiles, and by making sure the pages that hold them are accessible to search and retrieval crawlers.