What is Guardrails? Plain-English Definition

In more detail

Guardrails are the safety checks wrapped around an AI — filters and rules that screen what goes in and what comes out, blocking harmful requests and catching dangerous replies before they reach you. They sit around the model, separate from its own judgment.

Think of bumpers on a bowling lane: they don’t make the bowler skilled, they just stop the worst gutter balls. It matters because raw models will cheerfully follow terrible instructions; guardrails are why public chatbots mostly don’t. Jailbreaks and prompt injection are attempts to slip past them.

📌 See it in action

A user asks a chatbot for instructions to make a weapon. Before the model even sees the message, a guardrail flags it and returns a refusal. Another guardrail watches the output side: if a reply accidentally includes someone’s phone number, it gets caught and scrubbed on the way out.

Goes with

Alignment →Jailbreak →System Prompt →

Now put the vocabulary to work.

Meet the models →Which AI is mine?