Agentic AI Security in One Sentence (and Why Boards Care) скачать в хорошем качестве

Agentic AI Security in One Sentence (and Why Boards Care)

If you remember one sentence from this video, make it this: An agent is software that can be steered by text into taking actions. That’s why agentic AI security isn’t a vibes conversation about “model safety.” It’s system security. In this short episode, I explain agentic AI security in board-level terms: 'What makes an AI system agentic' 'Why the security boundary shifted from the model to the action supply chain' 'How a single malicious snippet can trigger real-world impact' 'What “good” controls actually look like in practice' This framing is designed for executives, board members, and security leaders who need clarity without jargon. Next episode: Prompt injection explained as a trust-boundary problem, not a “prompt problem.” References NIST AI 600-1: Generative AI Profile https://nvlpubs.nist.gov/nistpubs/ai/... OWASP AI Agent Security Cheat Sheet https://cheatsheetseries.owasp.org/ch...