Pentesting Index
Home / Glossary / AI Red Teaming

AI Red Teaming

Updated November 18, 2026Written by Dani N.Reviewed by Shir A.Methodology
Definition

AI red teaming is adversarial testing of AI systems, such as models, LLM applications and agents, to find ways they can be manipulated or abused.

What Is AI Red Teaming?

AI red teaming attacks an AI system the way an adversary would. It covers prompt injection, jailbreaks, data leakage, unsafe outputs and agents taking actions they should not. The most useful tests follow an attack from the prompt into the tools, APIs and data the AI can reach.

Types of AI Red Teaming

AI Red Teaming Use Cases

Limitations of AI Red Teaming

Model-only testing misses risks in the surrounding application. Results also vary between runs because model behavior is not deterministic, so continuous or repeated testing matters.

Related ranking

Related terms