---
title: "Red teaming"
description: "Structured adversarial testing of a model or agent to elicit failures (jailbreaks, injection, tool misuse) before an attacker does; treated here as an evidence-producing control."
canonical: https://aigovernanceengineer.com/glossary/red-teaming
author: "Jorge García Aibar"
license: "CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)"
doi: https://doi.org/10.5281/zenodo.22956197
version: "0.5.0"
updated: 2026-09-24
---

# Red teaming

Structured adversarial testing of a model or agent to elicit failures (jailbreaks, injection, tool misuse) before an attacker does; treated here as an evidence-producing control.

- Developed in: [ch. 05, Pattern: Adversarial Red-Team Suite](https://aigovernanceengineer.com/bok/patterns#pattern-adversarial-red-team-suite)
- Chapters: [ch. 04, The Stack](https://aigovernanceengineer.com/bok/the-stack) · [ch. 05, Patterns](https://aigovernanceengineer.com/bok/patterns) · [ch. 15, Deployment](https://aigovernanceengineer.com/bok/governing-deployment)
- In the glossary chapter: https://aigovernanceengineer.com/bok/glossary#t-red-teaming

## Sources

Defined by this Body of Knowledge: the section it is developed in is the source.