Hallucination
When a model confidently states something false.
Why it matters
Hallucination is the biggest trust problem in AI today. When a model confidently states something that is completely made up - fake citations, invented statistics, non-existent products - that is a hallucination.
Understanding this risk is critical before relying on AI for research, legal work, medical information, or any high-stakes decision. Tools with built-in fact-checking or source citations help reduce this risk.
How it works
3 stepsRelated terms
From the glossaryFrequently asked questions
Can hallucinations be eliminated?+
Not entirely with current architectures. They can be reduced with grounding techniques like RAG, output verification, constrained decoding, and human review, but a zero-hallucination guarantee does not exist yet.
Why do LLMs hallucinate?+
Models are trained to produce plausible next tokens, not to verify factual accuracy. When the correct answer is uncertain or absent from training data, the model generates a confident-sounding guess.
How can I detect hallucinations?+
Check outputs against authoritative sources, use a second model to critique the first, or implement retrieval so claims can be traced back to a cited document.
See the tools that use it.
The fastest way to understand Hallucination is to see it inside real products. Browse hand-reviewed tools that put it to work, each one checked by a person before it was listed.
Browse hand-reviewed AI tools