---
term: "Red-teaming"
category: "AI risks and safety"
url: https://agenticopenfinance.com/glossary/red-teaming
source: AgenticOpenFinance glossary
---

# Red-teaming

> Structured adversarial testing of an AI system to find flaws, harmful behaviours and vulnerabilities before attackers do.

AI red-teaming is a structured testing exercise in which a team probes an AI system, often adversarially, to find flaws, vulnerabilities and harmful or unexpected behaviours. NIST's Generative AI Profile describes it as an evolving practice for identifying potential adverse behaviour or outcomes of a model or system, how they could occur, and for stress-testing safeguards. For agents, red-teaming should target actions as well as words, for example trying to make the agent pay a new payee, exceed a limit or skip an approval.

Related terms: [Model validation](https://agenticopenfinance.com/glossary/model-validation.md), [Prompt injection](https://agenticopenfinance.com/glossary/prompt-injection.md), [Jailbreak](https://agenticopenfinance.com/glossary/jailbreak.md), [Excessive agency](https://agenticopenfinance.com/glossary/excessive-agency.md)