protologue

Guardrails

Also called safety filters, input/output filtering, safeguard model, moderation, content filter.

Guardrails are checks placed around a model call that screen inputs and outputs, using classifiers, rules, or a second model, to block, redact, or redirect content that violates a policy before it reaches the model or the user.

Description

Typical guardrails detect jailbreak attempts and prompt injection, unsafe content, personal data, and off-topic requests. Safeguard models such as Llama Guard classify messages against a taxonomy of risks, and toolkits such as NeMo Guardrails define programmable rails for dialogue. Guardrails add latency and can over-block, and none is complete, so they are layered with model training and least-privilege design.

Sources

  1. Inan et al. (2023). Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations.
  2. Rebedea et al. (2023). NeMo Guardrails: A Toolkit for Controllable and Safe LLM Applications with Programmable Rails.

Cite this entry

Protologue. (2026). Guardrails. In Protologue: A Taxonomy of Prompting and LLM Techniques (v1.4.0, PTL-0160). https://protologue.com/t/guardrails/

BibTeX
@misc{protologue_guardrails,
  title = {Guardrails},
  author = {{Protologue}},
  year = {2026},
  howpublished = {Protologue: A Taxonomy of Prompting and LLM Techniques, v1.4.0},
  note = {Entry PTL-0160},
  url = {https://protologue.com/t/guardrails/}
}

Markdown JSON