AI agents are moving from pilot to production—but enterprise security teams are still catching up. In this conversation, Box security leaders break down what agent security really means, why the risks are different from traditional AI assistants, and what organizations need to do now to protect their content and business. The discussion covers major enterprise AI risks including prompt injection, agent drift or misalignment, overly broad permissions, shadow AI, and the need for runtime and behavioral guardrails. It also outlines practical actions security leaders can take today: inventory what’s in use, prioritize controls based on risk, improve auditability and observability, and ask vendors tougher questions about how their agents actually operate. You’ll also hear why AI agent security requires a mindset shift—not just new controls—and how Box fits in as a secure content layer for enterprise AI. In this video: Why AI agents create higher-stakes security risks than assistants What prompt injection looks like in enterprise content How agent drift and misalignment can impact workflows Why traditional security models are not enough The importance of runtime security and behavioral guardrails How to handle shadow AI and unauthorized agent usage Three practical priorities: inventory, risk-based controls, and auditability What to ask AI vendors about agent security Why secure content is foundational to enterprise AI strategy If you’re exploring how to deploy AI agents safely at scale, this session offers a practical framework for security and IT leaders. #AIAgentSecurity #EnterpriseAI #Cybersecurity #AISecurity #PromptInjection #ShadowAI #CISO #BoxAI #DataSecurity #AICompliance

 23      0