Red Team AI

Definitions

1. The practice of adversarially testing AI systems to uncover unsafe behavior, security weaknesses, misuse pathways, or policy failures before deployment. Red teaming is a common safety practice for high-risk AI features.

   In plain English: Adversarial testing of an AI system to find weaknesses before release.

   Example: "They ran a red team AI exercise to probe prompt injection, unsafe advice, and data leakage before expanding access."

Related Terms

• Red Teaming AI
• AI Red Team
• AI Sandbox
• Interpretability
• Jailbreak AI
• LLM Security
• Prompt Attack
• Prompt Guard
• Refusal Training
• Safety Benchmark