Definition

An AI hallucination is the production of a factually false piece of information presented with confidence by a generative AI model. LLMs do not distinguish true from false: they statistically predict the next token without systematic access to a reliable source.

Why it is a brand risk

• Wrong information about products, prices, features
• Mentions of fictional partnerships or non-existent people
• Fake quotes attributed to executives
• Reputational risk when users treat AI output as truth
• Legal exposure in case of unintended defamation

Notable examples

Lawyers sanctioned for citing case law invented by ChatGPT, products described with fictional features by e-commerce chatbots, fake biographies generated about public figures.

How to protect yourself

• Actively monitor what AI says about your brand (AI citations)
• Strengthen EEAT signals and knowledge graph presence
• Publish official structured information through Schema.org markup
• Document and report critical hallucinations to vendors (OpenAI, Anthropic, Google)
• Prepare an internal AI crisis management workflow