Safety & Ethics
Glossary
Hallucination
When a model confidently states something that is false or unsupported by its sources.
Last updated
A hallucination is fluent but incorrect output: a made-up citation, a wrong statistic or an invented API. It happens because models predict likely text rather than verify facts. Grounding answers in retrieved sources, asking for citations and adding human review all reduce the risk, but no current system eliminates it entirely.
Related terms
Retrieval-Augmented Generation
A technique that lets a model look up relevant documents and include them in its answer, improving accuracy and freshnes...
Benchmark
A standard test used to compare model performance on a defined task.
Guardrails
Rules and checks that keep an AI system within safe, intended behaviour.
Keep exploring
Browse the full AI glossary or compare AI tools that use this technology.