John Willis: The Real Risk of AI Agents and the Future of DevOps Security скачать в хорошем качестве

John Willis: The Real Risk of AI Agents and the Future of DevOps Security

At SCALE 23 in Pasadena, Barton George interviews John Willis, co-author of The DevOps Handbook, about the rise of AI agents, rogue behavior, and the operational risks of autonomous systems. Willis argues that AI agents fundamentally change the risk model of software. Traditional security and operational assumptions were built around deterministic software and human oversight. With agents capable of spawning sub‑agents, executing multi‑system actions, and pursuing goals independently, risk grows exponentially with scale, autonomy, and velocity. Willis explains that the companies that succeed will be the ones that develop clear frameworks for agent autonomy, authority, and safety. Topics include: Rogue AI agents Deterministic vs probabilistic systems Shadow AI and enterprise risk Agent governance and security models Why proof-of-concept AI isn’t production-ready