Security controls built on static access policies designed for predictable behavior are colliding with systems that reason instead of simply executing. AI agents, driven by outcomes rather than rules, ...