Securing AI Agents in the Enterprise: Practitioner Lessons from the PBC AI Center of Excellence

As enterprises move from experimenting with AI agents to deploying them in real workflows, security teams are facing a new set of questions: What should agents be allowed to do? How do we govern permissions and blast radius? How do we detect shadow agents? Which frameworks are useful, and where are the gaps?

Over the past several months, the PBC AI Center of Excellence (COE) | Building Secure AI Agents cohort has been discussing these questions through practitioner-led working sessions. This virtual session shares key lessons, open questions, and real-world approaches from the cohort, including early insights from "Securing AI Agents in the Enterprise: A Practitioner’s Guide", linked below.
‍

𝗧𝗼𝗽𝗶𝗰𝘀 𝗰𝗼𝘃𝗲𝗿𝗲𝗱:

• Key security considerations for enterprise AI agents
• How practitioners are thinking about agent tasks, delegation, and permissions
• The current landscape of AI security frameworks and where they help
• Cost, governance, and operational trade-offs teams are working through
• Open questions the community is still exploring around agentic AI security