Guardrails

Safety mechanisms and filters applied to Generative AI systems to restrict harmful inputs, block toxic outputs, ensure policy compliance, and reduce hallucinations.

Part of the Data & AI Terms glossary.