Hallucination

Hallucinations can occur when generative AI models produce incorrect facts, invented citations, nonexistent events, or other unsupported claims. Because language models generate responses by predicting likely sequences of words, they may sometimes produce convincing information without verifying whether it is true. Hallucinations can also result from incomplete context, ambiguous prompts, or limitations in training data.

Hallucinations can reduce the reliability and trustworthiness of AI-generated information. In areas such as healthcare, finance, legal services, and enterprise decision-making, inaccurate outputs may create significant risks. Techniques such as grounding, retrieval-augmented generation, validation, and human review can help reduce their impact.