The Enterprise Guide to Agentic AI Risk: Security, Governance, and the Cross-Functional Playbook
Overview
| Experience | In Person |
|---|---|
| Track | Data Strategy |
| Industry | Enterprise Technology, Manufacturing, Financial Services |
| Technologies | Unity Catalog, Databricks Agents, Lakewatch |
| Skill Level | Intermediate |
When AI systems shift from analyzing to acting, the threat model changes fundamentally. Most enterprises are not ready. Agentic AI amplifies both the speed of innovation and the magnitude of risk, creating new attack surfaces and governance gaps that no single team can close alone. The organizations getting this right are building cross-functional response models that unite security, data, legal, and technology leadership around a shared risk framework. In this session, Databricks CISO Omar Khawaja introduces the lethal trifecta of agentic risk, the three converging conditions that create enterprise-scale exposure and the organizational playbook for neutralizing it. You'll walk away with a prioritized set of actions your enterprise can begin implementing in the next 30 days, and a model for the cross-functional partnerships that make agentic AI governable at scale.
Full Summary
Managing agentic AI risk: a framework for moving beyond the pilot
Enterprise AI is at an inflection point. The share of enterprise apps embedding agents is projected to jump from 5 percent to 40 percent in a year, and security buyers increasingly expect AI features by default. Yet most organizations still circle in pilots: 88 percent use AI, two thirds remain stuck in proof-of-concept mode, and only about one fifth report mature governance for autonomous agents. The gap between enthusiasm and production is where agentic risk concentrates.
FAQ
It is the simultaneous presence of three conditions: access to sensitive data, exposure to untrusted inputs, and the ability to act externally or change another system's state. Eliminating any one prevents the risk from being realized. For higher-stakes use cases, mitigating two adds further protection.
No. The framework is designed to be filtered against your architecture, deployment model, and use case. Organizations identify only the subset that applies, then select from the mapped 73 controls to address those specific risks.
Agentic systems act on behalf of users or departments, introducing about 40 risks beyond those in traditional ML or generative AI. The changes concentrate in agent memory, planning, the underlying model, and components such as MCP servers and clients.
DASF focuses on technical risk, component-level threats, and mapped controls. DAGF covers non-technical governance, including operating model, regulatory alignment, ethics, and transparency, with 43 considerations. Together they support a complete governance approach.
Enterprises often skip the orient and decide phases, jump straight to deploying broad controls, and attempt to solve the largest possible problem. A focused threat model and architecture-specific filtering make the scope smaller and the path to production faster.