Glossary

Guardrail

A control placed around a model that checks or constrains its inputs and outputs, such as a prompt-injection classifier, a content filter or a schema validator. Guardrails reduce risk probabilistically and can be bypassed, so authorization for consequential actions must still be enforced outside the model.

Defined in: OWASP Top 10 for LLM Applications 2026

Covered in

Guide

Contain prompt injection at the application boundary

Assume retrieved content can influence model output; enforce data access and consequential actions in components the model cannot authorize for itself.