AI Red-Teaming is a structured, adversarial testing process where security experts (the "Red Team") actively and systematically attack or misuse an Artificial Intelligence system (especially Large Language Models) to uncover vulnerabilities. The goal is to identify and exploit weaknesses related to safety, security, and fairness, such as jailbreaks (bypassing safety guardrails), prompt injection, data leakage, or the model generating harmful or biased content, before malicious actors can exploit them. It is a critical component of proactive AI risk management.

‍