Glossary
Red Teaming (AI)
Last reviewed
Structured adversarial testing of an AI system, where evaluators deliberately try to elicit unsafe, biased, or unintended outputs to surface risks before deployment. A required practice for foundation models under the EU AI Act.