[ IA Inside #12] Quand l'IA attaque l'IA : AI Red Teaming

Welcome to a new episode of AI Inside. AI-powered chatbots and autonomous agents are rapidly becoming part of everyday business operations. As these systems gain access to sensitive data and critical workflows, one question becomes increasingly important: how do you know they'll behave as intended? Traditional security testing was never designed for systems that understand and generate natural language. That's exactly where AI Red Teaming comes in. In this episode, Michael Molho explains why testing AI requires a fundamentally different approach and how AI can be used to identify weaknesses in other AI systems before they reach production. Chapters 00:00 - 01:13 Intro 01:13 – 03:02 Two families, two different risks 03:02 - 05:17 Why pentesting is no longer enough 05:17 - 07:37 AI Red Team agent 07:37 - 09:48 Conclusion What you'll learn : Why AI chatbots and AI agents introduce different security risks. Why traditional pentesting reaches its limits with AI. How an AI Red Team agent works. Why AI has become the most effective way to test another AI. How continuous AI Red Teaming strengthens security over time. Watch the latest AI Inside episode on our YouTube channel. Like, comment and share if this episode helps you better understand the emerging security challenges of artificial intelligence.