Guardrails
Working knowledgeGovernance, Safety and Ethics
Guardrails are the rules and filters built around an AI system that constrain what it can produce, access or do. They operate in two directions: input guardrails filter what users are permitted to ask, and output guardrails filter what the model is permitted to say. Together they enforce topic restrictions, keep sensitive data out of responses and hold outputs to a required format.
In practice
Guardrails are the difference between an AI deployment and a governed one, and they are usually the first thing an auditor or an insurer asks to see. They are also the cheapest control to add and the easiest to under-specify, so decide before launch what the system must never disclose and what it must never be allowed to decide.