Skip to content

Red-teaming

Adversarial testing of an AI system by a team simulating attackers or misusers to find failure modes before they become incidents.

Adversarial testing of an AI system by a team simulating attackers or misusers to find failure modes before they become incidents.

Related terms


← All glossary terms