AI engineering glossary · Evals guardrails AI security
What is output guardrail?
Output guardrail: A check applied to a model's reply before it is shown or acted on, which can block, redact or regenerate it. Typical checks cover toxic language, leaked personal data, unsupported claims and invalid formats.
Output guardrail: an example
A filter scans each reply for phone numbers and masks them before the answer is displayed.
Related terms
More Evals guardrails AI security terms
- Over-refusal
- Overreliance
- OWASP Top 10 for LLM Applications
- Pairwise comparison
- Payload splitting
- Perplexity filter
- Pickle deserialization risk
- PII detection
Where you learn it
Output guardrail is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.