An offensive security testing/assessment process that involves having a group of people (the "red team") simulate an adversarial attack to penetrate or corrupt the target, which could be artificial intelligence systems or models, policies, or assumptions; used to identify vulnerabilities, demonstrate the effect of a potential attack, or test the strength of defenses. See Blue Teaming and Purple Teaming
