NeMo Guardrails is NVIDIA's open-source toolkit for adding programmable rules, called rails, around an LLM chatbot to keep it on topic and safe.
In plain words
NeMo Guardrails sits between your application code and the language model. You write rails that control what the model may do: avoid certain subjects, answer specific requests in a set way, follow a fixed dialogue path, keep a particular style or pull out structured data. It also includes protections against common attacks such as jailbreaks and prompt injection.
A quick example
Imagine a bank's support bot. With NeMo Guardrails, the team adds a rail so that if a user asks for investment tips or political opinions, the bot politely declines and steers back to account questions.
Where you'll meet it
- Customer-facing chatbots that must stay on topic.
- Adding jailbreak and prompt-injection defences to LLM apps.
Related
Sources
- DocsNeMo Guardrails README, NVIDIA (GitHub) · read 29 Sept 2026
- PaperNeMo Guardrails: A Toolkit for Controllable and Safe LLM Applications with Programmable Rails, Rebedea et al. (arXiv) · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.